The objective of this study was to compare the optimal replacement decisions using two alternative state-of-the-art models: the optimization dynamic programming model and the Markov chain simulation model. Lactation, month in milk and pregnancy status were used to describe cow states in a herd in both models. Both models were fed with the same parameters and transition probabilities to make the fairest comparison possible. The cow value calculated by the Markov chain model was compared against the retention pay-off estimated by the dynamic programming model. These values were used to rank all the animals in the herd. Then, the rank correlation (Spearman’s correlation) was calculated between results of both models. The overall correlation was 95%, which showed a strong linear relationship between rankings of animals from the two models. Moreover, the lowest 10% ranking cows -which are the most likely replacement candidates- displayed a greater correlation, 98%. Thus, the final replacement decisions with both models were similar. A post optimality analysis was used to explore the effect of the optimal replacement decisions on the herd dynamics and herd net return. The results showed a comparable herd structure by both models. A net return was improved US$6/cow per year by using replacement decisions of both dynamic programming model and the Markov chain cow value model.