Altman's Z-score has been used for several decades to calculate bankruptcy probability. However, the conventional Z-score fails to consider possible earnings manipulations that could change the fundamental accounting figures and their implications for investors' decision models. We reconstruct the Z-score, making adjustments for earnings management.We apply the adjusted Z-score to measure the degree of deviation from bankruptcy probability for the bankruptcy sample. We find that the Z-score is overstated (respectively, understated) for the income-increasing (respectively, income-decreasing) earnings-management sample. Furthermore, we find that the adjusted Z-score performs better than the Z-score for bankruptcy predictions.