Several confounding factors may influence the outcome of an experiment and the extent of inter-individual variation. The aim of this study was to investigate if cage enrichment induces an effect on experimental mean values and on inter-individual variation in the light/dark paradigm using diazepam as the anxiolytic drug. The behaviour of 216 naive adult male mice of two different strains (BALB/c and C57BL/6) was studied. The animals were housed in groups of four in 'non-enriched', 'enriched' (nesting material) or 'super-enriched' (nest-box, nesting material, wooden gnawing stick and PVC tube) cages. After 5 weeks the animals were assigned to one of three treatments: control (no injection), sham (saline injection i.p.) or diazepam (1 mg/kg bw i.p.) and tested in the light/dark test for 5 min. Variation data were analysed using three different methods (mean absolute deviation, coefficient of variation and power analysis). The C57BL/6 mice scored higher than BALB/c mice in activity related measurements and showed a less 'emotional' behaviour profile in the pharmacological control situation of the light/dark test. In this study the anxiolytic effect of diazepam was clear in BALB/c mice but absent in C57BL/6 mice. Mice housed in enriched and super-enriched cages gained more weight than mice in non-enriched cages, although food intake was not affected. Generally, the strain of mouse had the greatest impact on both mean values and variation. However, there was no consistent increase for one particular strain. The choice of statistical method for analysing variation may influence the interpretation of within-group variability, but none of the methods showed any significant differences between standard and enriched conditions on variability in any of the parameters measured.