Medication calculation errors occur in all settings and can cause serious disruption in treatment protocols. This experimental study used a pretest-posttest, control group design. Sixty-seven RNs from three healthcare agencies completed all phases of the study. All nurses completed a 20-item medication calculation test, used both as the pretest and the posttest, and completed a short questionnaire that contained demographic and attitudinal items. After taking the pretest, nurses were assigned randomly to one of three experimental groups or a control group. Intervention times for the experimental groups totaled 3 hours, followed by a posttest 4 to 5 months after the pretest. Scores on the pretest ranged from 25% to 100% (mean = 75.5%, standard deviation = 15.13) and from 30% to 100% on the posttest (mean = 80%, standard deviation = 14.72). Scores improved for all groups except the self-study workbook group. No significant difference was found between the experimental groups or the control group for posttest medication calculation test scores. The classroom intervention was most costly, and the workbook intervention was least costly. The nurses identified the workbook intervention as the most satisfying method and computer-assisted instruction (CAI) as least satisfying. A strong positive correlation existed between the nurses' self-assessment of comfort and skill levels with medication calculation test scores. Questions necessitating multiple calculations and those necessitating a conversion not provided were the most difficult to answer correctly. Staff development educators need to address their role in improving the effectiveness of medication calculation.