Objective: To determine the reason for the low response to the exclusive use of prostaglandin in synchronization programs in cows under tropical grazing compared with the use of progesterone (CIDR). Design/Methodology: Thirty-five cows with CL were randomly distributed in two groups. The first group (GPG; n=23) was synchronized using two doses of PGF2α (25 mg of Dinoprost®) with a 12-day interval. The second group (GCIDR; n=12) was synchronized with an intravaginal device (1.9 g of progesterone, 2 mg of estradiol benzoate, and 50 mg of progesterone); a PGF2α (pm) dose was applied on day 7, before removing the CIDR (am) on day 8. CL regression, luteolytic failure, progesterone concentration, and CL size were determined. Data was subjected to a normality test, followed by the Mann-Whitney U test or independent Student's t-test and chi-square test. Results: Only 82.6 % (19 out of 23) of the GPG cows that received the second dose of prostaglandins have a functional CL (<1 ng mL-1 of progesterone). The CL recorded a regression only in 43.5% of the cows in GPG vs. 91.7% in GCIDR (P=0.0001). In addition, GPG cows showed a luteolytic failure of 39.1% and an asynchrony of 17.4%. Conclusions: The low effectiveness of prostaglandin on the synchrony and regression of the CL (luteolytic failure) in cows fed under tropical grazing can be attributed to the low efficiency of the synchronization programs.