RMR vs Q-System: A Complete Comparison Guide for Engineers
The Bieniawski Rock Mass Rating (RMR) system and the Barton Q-System are the two dominant rock mass classification methods used in geotechnical practice today. Both convert field observations of rock conditions into numerical indices that drive support design, excavation strategy, and construction risk management. Yet they take fundamentally different mathematical approaches, weight different parameters, and behave differently across the spectrum of rock quality. This guide compares the two systems side by side so engineers can choose the right tool for the job — or better still, use both. For hands-on work, you can run the numbers in our RMR calculator and our Q-System calculator on the same dataset and compare results directly.
Quick Comparison at a Glance
Before diving into the technical details, the table below summarises the key structural differences between RMR89 and the Q-System. Each row reflects a design decision made by the original authors that has consequences for how the system performs in practice.
| Feature | RMR (Bieniawski 1989) | Q-System (Barton 1974) |
|---|---|---|
| Developer | Z.T. Bieniawski (CSIR, South Africa) | N. Barton, R. Lien, J. Lunde (NGI, Norway) |
| Mathematical structure | Additive (sum of 6 parameters) | Multiplicative (3 ratios) |
| Output range | 0 to 100 (linear) | 0.001 to 1000 (logarithmic) |
| Number of parameters | 6 | 6 (grouped into 3 ratios) |
| Stress effects | Not directly included | Stress Reduction Factor (SRF) |
| Intact rock strength | Yes (UCS, up to 15 points) | Not directly (only via SRF) |
| Calibration database | Civil tunnels and mines | 200+ Norwegian hard-rock tunnels |
| Best application | Civil works, slopes, foundations | Hard-rock tunnelling, deep excavation |
| Original publication | 1973, updated 1989 | 1974, updated 1993, 2002 |
Origins and Development History
The two systems were developed independently in the early 1970s on different continents to solve closely related problems. Zbigniew Bieniawski published the first version of RMR in 1973 while working at the South African Council for Scientific and Industrial Research (CSIR). His original database came from coal mining and civil tunnelling case histories in southern Africa. Bieniawski revised the system in 1976, 1979, and most importantly in 1989, when he published the comprehensive monograph Engineering Rock Mass Classifications. The 1989 revision introduced the five sub-parameters for joint condition, refined the groundwater rating, and provided separate orientation adjustment tables for tunnels, slopes, and foundations. The 1989 version is the one used in nearly all modern RMR calculators, including this site.
The Q-System emerged a year later from the Norwegian Geotechnical Institute (NGI). Nick Barton, Reidar Lien, and Jan Lunde published their original paper in 1974 after analysing more than 200 case histories of Scandinavian hard-rock tunnels. Norway's geology — predominantly competent crystalline basement rocks at substantial depth — shaped the system's emphasis on jointing geometry and stress conditions. The Q-System has been periodically updated, most notably in 1993 with revised SRF values to better capture squeezing ground, and again in 2002 with the introduction of Qtbm for tunnel boring machine performance prediction. For background on why both systems exist and how they fit into the broader landscape, see our guide on what Rock Mass Rating is.
Mathematical Structure: Additive vs Multiplicative
The most important conceptual difference between the two systems is their mathematical structure. RMR is purely additive: each of the six parameters is rated independently, and the ratings are summed to produce a total score from 0 to 100. The maximum contribution of each parameter is fixed: UCS contributes up to 15 points, RQD up to 20, joint spacing up to 20, joint condition up to 30, groundwater up to 15, and the orientation adjustment subtracts up to 60 points depending on application type. The additive structure has clear advantages: it is intuitive, easy to teach, and easy to audit because each parameter contribution is visible in the final score.
The Q-System uses a multiplicative formula:
Q = (RQD / Jn) × (Jr / Ja) × (Jw / SRF)
where RQD is Rock Quality Designation, Jn is the joint set number, Jr is the joint roughness number, Ja is the joint alteration number, Jw is the joint water reduction factor, and SRF is the stress reduction factor. Each of the three ratios captures a distinct aspect of rock mass behavior: (RQD/Jn) represents block size, (Jr/Ja) represents inter-block shear strength, and (Jw/SRF) represents the active stress condition. Because the formula multiplies these ratios together, the output spans roughly six orders of magnitude — from 0.001 (exceptionally poor squeezing ground) to 1000 (massive unjointed rock).
The practical consequence is sensitivity at the extremes. In very poor rock, RMR compresses many real-world conditions into the narrow band of 0 to 20 points, making it hard to discriminate between merely difficult ground and impossible ground. The Q-System spreads the same range of conditions across two full log decades (0.001 to 0.1), preserving meaningful distinctions where they matter most for safety. Conversely, in very good rock, both systems become less sensitive because support requirements collapse to "no support needed" regardless of the exact numerical value.
Parameter-by-Parameter Comparison
Although both systems aim to characterize the same physical rock mass, they use different parameters and weight them differently. The table below maps the parameters of each system to their counterparts in the other.
| What it measures | RMR parameter | Q-System parameter |
|---|---|---|
| Intact rock strength | Parameter 1: UCS (0–15 pts) | Not included directly |
| Degree of fracturing | Parameter 2: RQD (3–20 pts) | RQD (numerator of first ratio) |
| Joint set count and pattern | Parameter 3: Joint spacing (5–20 pts) | Jn (joint set number) |
| Joint surface character | Parameter 4: Joint condition (0–30 pts) | Jr (roughness) and Ja (alteration) |
| Groundwater conditions | Parameter 5: Groundwater (0–15 pts) | Jw (joint water reduction) |
| Excavation orientation | Parameter 6: Adjustment (-60 to 0) | Not included; handled by user |
| In-situ stress | Not included | SRF (stress reduction factor) |
Two important observations follow from this mapping. First, RMR explicitly accounts for intact rock strength through UCS, while the Q-System assumes rock strength is captured indirectly through other parameters. This makes RMR more sensitive to weak rock conditions where the intact strength itself is the limiting factor — for example, weathered shales, mudstones, or marls. The Q-System can underestimate the difficulty of these materials unless the user is careful with SRF selection.
Second, the Q-System explicitly addresses stress through SRF, while RMR is silent on the topic. For deep tunnels (typically deeper than 300 to 500 meters in moderately strong rock) or tunnels in weak rock at any depth, stress-induced behaviors such as spalling, slabbing, and squeezing become important. The Q-System captures these through SRF values that can dramatically reduce Q (down to 0.5 for moderately stressed massive rock or 5 to 20 for severe squeezing ground). RMR users facing these conditions need to apply separate stress checks using methods like the Hoek-Brown criterion, the convergence-confinement method, or numerical modelling. For step-by-step RMR calculation including the orientation adjustment, see our guide on how to calculate RMR.
Correlations Between RMR and Q
Because the two systems are widely used and engineers often need to translate between them, several empirical correlations have been published over the years. The most cited is Bieniawski's original 1976 correlation, derived from 117 case histories where both systems had been applied to the same rock masses:
RMR = 9 ln(Q) + 44
This equation gives a one-to-one mapping. For Q = 1, RMR = 44. For Q = 10, RMR = 65. For Q = 100, RMR = 85. For Q = 0.1, RMR = 23. The equation works reasonably well in the central range of rock quality but breaks down at the extremes, where one system becomes insensitive while the other still varies. Bieniawski himself reported a standard deviation of about 18 RMR units around the regression line, meaning that for a given Q value, the actual RMR could differ by plus or minus 18 points in any specific case.
Other correlations have been proposed for specific geological settings. Rutledge and Preston (1978) developed RMR = 5.9 ln(Q) + 43 for New Zealand tunnelling cases. Moreno (1980) proposed RMR = 5.4 ln(Q) + 55.2. Cameron-Clarke and Budavari (1981) developed RMR = 5 ln(Q) + 60.8 for deep South African mining excavations. Goel et al. (1996) studied 63 Indian Himalayan tunnel cases and proposed an alternative form using rock mass number N rather than Q to remove the influence of SRF, finding much better agreement when stress effects were excluded from the comparison.
The practical lesson from these correlations is clear: do not use them as a substitute for direct calculation. They are useful for sanity checks, for converting historical data from one system to another for database compilation, and for benchmarking unusual results. But they should never replace running both systems independently when both are being used on a project.
When to Use RMR
RMR is the better choice for several common engineering applications. For shallow civil tunnels in moderate to good rock, where the orientation of joints relative to the excavation is the dominant concern, RMR's explicit orientation adjustment provides clear guidance on how the rock mass score should be modified. For rock slopes, the RMR-based Slope Mass Rating (SMR) developed by Romana in 1985 offers a structured framework that is widely used in slope engineering practice. For foundation design on rock, the RMR foundation orientation table provides direct support for evaluating rock-rock interface conditions under spread footings, drilled shafts, and dam abutments.
RMR is also preferred in educational settings because its additive structure is intuitive and the parameter contributions are transparent. Students can see how each parameter affects the total score and develop a sense for which conditions matter most. For projects in regions where civil tunnelling tradition follows the British, American, or Australian practices, RMR is more likely to be the lingua franca among engineers, contractors, and regulators. National codes and design standards in countries including South Africa, India, Turkey, and many parts of South America reference RMR as the primary classification method.
When to Use Q-System
The Q-System is the better choice for hard-rock tunnelling at depth, where stress effects matter and where the rock quality may vary across many orders of magnitude. Norwegian and broader Scandinavian tunnelling practice has institutionalized the Q-System through the Norwegian Method of Tunnelling (NMT), which links Q values directly to support charts for fibre-reinforced shotcrete and rock bolts. Anyone working on tunnels in Scandinavia, or following Scandinavian design philosophy elsewhere, should default to Q.
The Q-System is also preferred for tunnel boring machine projects, where the Qtbm extension developed by Barton in 2002 provides explicit linkages between rock quality and TBM advance rates, cutter wear, and ground conditioning requirements. For deep underground research laboratories, deep mining excavations, and high-pressure water tunnels, the Q-System's sensitivity in the poor-rock end of the spectrum and explicit handling of stress make it more suitable than RMR.
Why Best Practice Uses Both
For important projects with significant geotechnical risk, the most defensible approach is to apply both systems independently and compare the results. When RMR and Q give consistent indications — both pointing to fair rock, for example — confidence in the classification is higher and design decisions can proceed with reduced uncertainty. When they diverge, the divergence itself is information: it usually highlights some aspect of the rock mass where the two systems weigh things differently, which is worth investigating before finalizing the design.
Running both systems also provides protection against the limitations of each. If the Q value seems unrealistically high because SRF was set too low, the RMR result acts as a sanity check. If the RMR value seems too high because the orientation adjustment was applied too leniently, the Q result raises a flag. Major tunnelling projects typically include both classifications in their geological logging templates, and many design specifications require both to be reported for each excavation face. For practical worked examples where both systems would be relevant, see our RMR worked examples page.
Limitations Common to Both Systems
Despite their differences, RMR and Q share several important limitations that any engineer should keep in mind. First, both are empirical systems calibrated against case histories from specific geological settings. Applying them to dramatically different rock types — for example, evaporites, swelling clays, or volcanic tuffs — should be done with caution and supplemented with site-specific testing. Second, both depend critically on the quality of input data. A poorly executed core run, an inexperienced field mapper, or a hastily selected RQD value can produce misleading results from either system.
Third, neither system is a substitute for engineering judgment or numerical analysis on critical structures. They provide first-order estimates suitable for preliminary design, contractual classification, and rapid construction-phase decisions, but final designs for high-consequence excavations should be based on a combination of classification, analytical methods, and numerical modelling. Fourth, both systems implicitly assume that the rock mass behaves as a continuum at the scale of the excavation, which may not hold true for very small openings (where individual joint blocks dominate) or very large openings (where regional structure becomes important).
Practical Recommendations
For an engineer choosing between RMR and Q-System on a new project, the following guidelines apply. If the project is a shallow civil tunnel, a road cut, a slope, or a foundation, default to RMR and supplement with Q if a second opinion is needed. If the project is a deep tunnel in hard rock, a TBM bore, or a Scandinavian-style underground excavation, default to Q and supplement with RMR. If the project is large, novel, or carries significant risk, plan to use both from the outset and budget the additional logging time accordingly.
Whatever system you choose, calibrate it against the local geology by comparing your classification results with the actual ground response observed during construction. Both Bieniawski and Barton emphasize that classification systems are tools for transferring experience between projects, and that experience is only useful if the engineer continually updates their understanding of how classification translates into real ground behavior at the specific site. To go deeper into either system, see our complete RMR guide or use the Q-System calculator to run a parallel calculation on the same dataset.
Frequently Asked Questions
Neither system is universally better. RMR is more intuitive due to its additive 0 to 100 scale and is easier to learn, making it widely used in civil tunnelling, slopes, and foundations. The Q-System is more sensitive in poor and very poor rock because of its logarithmic 0.001 to 1000 scale and integrates stress conditions explicitly through the SRF parameter. Best practice on important projects is to apply both systems independently and compare results, since each highlights different aspects of rock mass behavior.
The most cited correlation is Bieniawski's original 1976 equation: RMR = 9 ln(Q) + 44. This gives reasonable agreement for moderate rock conditions (Q between 0.1 and 100, RMR between 20 and 80) but breaks down at extreme ends. Other correlations include Rutledge (1978): RMR = 5.9 ln(Q) + 43, and Goel et al. (1996) for Indian Himalayan tunnels. All correlations have substantial scatter, typically plus or minus 18 RMR points around the regression line. They should be used only as cross checks, never as a substitute for direct calculation of both systems.
Barton selected a logarithmic scale because rock mass quality varies over many orders of magnitude in nature. Going from very poor squeezing ground to massive intact granite represents a strength and stiffness change of roughly four orders of magnitude. A linear additive scale like RMR compresses this range and loses sensitivity at the extremes. The Q-System's multiplicative formula naturally produces values from 0.001 (exceptionally poor) to 1000 (exceptionally good), giving uniform resolution across the full quality spectrum.
The Q-System explicitly incorporates stress through the Stress Reduction Factor (SRF), which adjusts for high stress in massive rock, squeezing or swelling ground, and shear zones. The Bieniawski 1989 RMR system does not include a direct stress parameter. For deep tunnels where stress effects dominate, the Q-System provides better discrimination. RMR users handling high-stress conditions typically apply additional stress checks separately, such as the Hoek-Brown failure criterion or numerical modelling, to supplement the basic classification.