There is a growing body of evidence that suggests strength training can be a safe and effective method of exercise for children. The medicine ball throw test can be used to measure children's muscular strength for the purpose of diagnosis and promotion of strength, but there is a lack of evidence of test score reliability. The purpose of this study was to establish the optimum combination of number of days and trials needed to obtain reliable estimates of medicine ball throw test scores. A total of 105 kindergarten children (45 girls and 60 boys; ages 5-7 years old) participated in the study. The participants were tested on two different occasions. On Day 1, the participant performed the medicine ball throw test and on Day 2, which was seven days after Day 1, the medicine ball throw test was repeated. Each participant sat on the floor with his/her back against the wall. The participants were asked to lift the medicine ball to their chest and throw it forward as far as they can. Each participant performed two practice throws, and the distance of the next three throws was recorded with a one minute rest period between each throw. The measurements were taken to the nearest 1/8 of an inch. Generalizability theory through a G-study was used to calculate the error associated with the distance of the medicine ball throw for each participant. Follow-up D-studies were performed to estimate reliability coefficients (G coefficients). The facets included were participants (P), number of days (D) and number of trials (T). The measurement model used was participants by days by trials (P x D x T). The G-study revealed that the participant facet accounted for 78% of the variance, the interaction between participants and days was responsible for 5%, and the interaction between participants, days, and trials was responsible for 17% of the variance. Days, trials and their interactions explained less than 1% of the variance. The D-studies showed that to achieve the desired reliability (i.e., G = .80 or greater), a combination of one day and two trials or two days and one trial are necessary (G = .85 and .87; respectively). The results from our investigation suggested that at least two trials or more be considered when utilizing the medicine ball throw test. Further investigation to establish validity evidence of the test is warranted.Keyword(s): assessment, measurement/evaluation, physical education PK-12