This study aimed to investigate the effects of different work and recovery characteristics on the W' reconstitution and to test the predictive capabilities of the W'BAL model. Eleven male participants (22 ± 3 yr, 55 ± 4 mL·kg⋅min) completed three to five constant work rate tests to determine CP and W'. Subsequently, subjects performed 12 experimental trials, each comprising two exhaustive constant work rate bouts (i.e., WB1 and WB2), interspersed by an active recovery interval. In each trial, work bout characteristics (P4 or P8, i.e., the work rate predicted to result in exhaustion in 4 and 8 min, respectively), recovery work rate (33% CP or 66% CP), and recovery duration (2, 4, or 6 min) were varied. Actual (W'ACT) and model-predicted (W'PRED) reconstitution values of W' were calculated. After 2, 4, and 6 min recovery, W'ACT averaged 46% ± 2.7%, 51.2% ± 3.3%, and 59.4% ± 4.1%, respectively (P = 0.003). W'ACT was 9.4% higher after recovery at 33% CP than at 66% CP (56.9% ± 3.9% vs 47.5% ± 3.2%) (P = 0.019). P4 exercise yielded a 11.3% higher W'ACT than P8 exercise (57.8% ± 3.9% vs 46.5% ± 2.7%) (P = 0.001). W'ACT was higher than W'PRED in the conditions P4-2 min (+29.7%), P4-4 min (+18.4%), and P8-2 min (+18%) (P < 0.01). A strong correlation (R = 0.68) between the rate of W' depletion and W' recovery was found (P = 0.001). This study demonstrated that both the work and recovery characteristics of a prior exhaustive exercise bout can affect the W' reconstitution. Results revealed a slower W' reconstitution when the rate of W' depletion was slower as well. Furthermore, it was shown that the current W'BAL model underestimates actual W' reconstitution, especially after shorter recovery.