Pedal claw geometry can be used to predict behaviour in extant tetrapods and has frequently been used as an indicator of lifestyle and ecology in Mesozoic birds and other fossil reptiles, sometimes without acknowledgement of the caveat that data from other aspects of morphology and proportions also need to be considered. Variation in styles of measurement (both inner and outer claw curvature angles) has made it difficult to compare results across studies, as have over-simplified ecological categories. We sought to increase sample size in a new analysis devised to test claw geometry against ecological niche. We found that taxa from different behavioural categories overlapped extensively in claw geometry. Whilst most taxa plotted as predicted, some fossil taxa were recovered in unexpected positions. Inner and outer claw curvatures were statistically correlated, and both correlated with relative claw robusticity (mid-point claw height). We corrected for mass and phylogeny, as both likely influence claw morphology. We conclude that there is no strong mass-specific effect on claw curvature; furthermore, correlations between claw geometry and behaviour are consistent across disparate clades. By using independent contrasts to correct for phylogeny, we found little significant relationship between claw geometry and behaviour. 'Ground-dweller' claws are less curved and relatively dorsoventrally deep relative to those of other behavioural categories; beyond this it is difficult to assign an explicit category to a claw based purely on geometry.