PURPOSE: To determine the level to which the creatine kinase (CK) response to damaging exercise can be predicted by other variables commonly used in the assessment of exercise-induced muscle damage. METHODS: 48 male participants, who were unaccustomed to eccentric exercise, performed 3 sets of 15 maximal lengthening contractions of the elbow flexors at 30°·s-l. CK activity in blood, muscle soreness (using the Talag scale), elbow range of motion, upper arm girth, elbow flexor isometric torque and isokinetic torque at 60°·s-l and 210°·s-l were measured pre-exercise and at 48 h and 96 h post-exercise. Participants were categorised as either LOW or HIGH responders according to their peak CK measurement (<750 and ≥750 IU·L-1, for LOW and HIGH, status respectively). Stepwise logistic regression analysis was performed with CK status as the dependent variable and with muscle soreness and the percent change in: range of motion (%ROM), upper arm girth, isometric torque (%ISOM), 60°·s-l isokinetic torque (%ISOK60), and 210°-s-l isokinetic torque (all at 48 h post-exercise) as predictor variables. RESULTS: 27 cases were classified as LOW (CK = 401±177 IU·L-1) and 21 cases as HIGH (CK = 1239±358 IU·L-1). The most parsimonious model included %ROM, %ISOM, and %ISOK60: log(p/(1-p)) = −1.963- (0.057 × %ISOM) + (0.067 × %ISOK60) - (0.149 × %ROM); where p is the probability of HIGH CK status. The model accounted for between 25% and 34% of the variance in CK status, with 85% of LOW and 67% of HIGH cases being correctly predicted; overall, 77% of predictions were accurate. CONCLUSION: This study supports previous literature reporting the high inter-individual variation in CK response to damaging exercise. Further, the results indicate that only some of the non-invasive measures commonly used to assess exercise-induced muscle damage may be used to predict the CK response to damaging exercise. Further research is required to elucidate the relationship between variables commonly used in the assessment of exercise-induced muscle damage.