Background and Aims: Tablets can be split by patients for a number of reasons, using various instruments. Tablets can be scored or unscored; if scored, they may be split into pieces, and in spite of guidelines to do so patients are still at risk of resultant drug dose fluctuations, or being exposed to toxic or subtherapeutic doses. The aim of this study was to investigate differences in weight between halves of tablets, split by different populations and with different devices. Methods: 3-factor full factorial design (3 runs) was used with participants: patients, caregivers, nurses, medical doctors, and pharmacists; instruments: scissors, tablet cutters, knives, hand; drugs: losartan, clonidine, metoprolol, and warfarin. The risk of unequal tablet splitting was estimated and analyzed for each factor and their interaction with linearized generalized models. Results: Differences in weight were found to be above 15% and 25% of the theor. weight as in general, the highest weight variations after splitting were found in clonidine with patients using scissors. The overall risk of non-equal tablet splitting was 22.5% for deviations > 15% and for > 25%. Conclusion: In this study, no tablet was split into halves of equal weight; based on these findings, splitting tablets is a questionable practice.