The usefulness of serum creatine kinase (CK) activity in the detection of Duchenne muscular dystrophy (DMD) carriers is dependent upon a reliable control distribution. In controls there is a small but reproducible seasonal variation in CK activity, with a statistically significant variation in the upper 95th percentile (78 IU/liter in May, 53 IU/liter in November, as compared with 67 IU/liter for the whole calendar year). Because CK values in DMD heterozygotes are higher in November than in May, the carrier detection rate may be highest in November. Failure to consider this seasonal variation in controls may cause misclassification of too many normal subjects. If tests are conducted throughout the year, however, the seasonal influence can be reduced by serial testing at intervals of several months. In this case, use of the highest of the results obtained in 3 tests compared with the normal range of single measurements misclassifies too many normal subjects. Use of the mean of 3 determinations provides more accurate classification.