The use of measures of strength variability as a means of determining sincerity of effort is becoming a more common practice, particularly in medico-legal and rehabilitation settings. The stability of such variability measures, however, has not been documented. This research investigated, in two studies, the trial-to-trial variability of grip strength under maximal and submaximal effort conditions. In the first study, 63 subjects were asked to give 100% grip effort, and in the second study 40 subjects from the original group were asked to give 50% grip effort. The Jamar hand dynamometer was used to measure grip strength in both experiments, and a coefficient of variation (CV) was calculated for every three repeat measures at each handle position. Testing was conducted on two separate occasions for both experiments. Although the interoccasion reliability of grip strength was very high, in comparison, the CV was not stable over test occasions, with interoccasion reliability indices close to zero. Factors significantly influencing CV were effort level, with submaximal effort producing larger CVs, and gender, with females having greater strength variability. If the rule is applied that one or more CVs above the 7·5% cut-off value could indicate submaximal effort, then for this sample of subjects giving maximal effort, 97% of females and 64% of males would be misclassified. Applying a single CV classification cut-off value to a mixed sample of subjects appears to unfairly discriminate against the females. Further research into the factors associated with high CV values is essential before the CV can be used with any confidence in a clinical setting as a method for determining sincerity of effort.