Several randomization procedures were tested to verify effects on diabetes-relevant traits (pancreatic insulin content (PIC), glucose tolerance (GT), incidence and age at onset of diabetes), which were calculated as mean +/- SD for each group. Estimation of PIC and GT were performed in LEW.1W/MaxK and BB/OK rats of our own inbred colonies. Furthermore, BB/OK rats were monitored for onset of diabetes. Significant differences between four (LEW.1W/MaxK and BB/OK for PIC) and 12 (BB/OK) litters and between groups arranged by several randomization procedures could be observed. Comparisons of first and second litters of the same parents in BB/OK rats suggest that, using these litters, significant differences between arranged groups can be excluded. Effects of randomization have to be considered in planning and performing experiments as well as in the interpretation of experimental results.