Our objectives were to characterize parameters of luteal activity based on milk progesterone concentration (P4c) data from before and after artificial insemination (AI) and to evaluate their potential association with fertility in Holstein cows. Records of AI events (n = 4,353) and of milk P4c (n = 158,961) obtained through an in-line milk analysis system (Herd Navigator, DeLaval International, Tumba, Sweden) from 1,891 lactations of 1,423 Holstein cows were evaluated. Milk P4c (ng/mL) were measured every 2.2 ± 1.9 d (mean ± standard deviation) between 23.6 ± 7.3 and 185.3 ± 56.7 d in milk. Variations in milk P4c of consecutive records were used to determine onset of luteal phase (increase in P4c from <5.0 to ≥5.0 ng/mL), luteal phase length (period, in days, of P4c ≥5.0 ng/mL), cessation of luteal phase (decline from ≥5.0 to <5.0 ng/mL, designated as P4c-decline), and pregnancy (AI followed by a luteal phase that remained uninterrupted until 50 d post-AI). The length of the luteal phase preceding AI, the highest P4c (P4c peak) during the luteal phase preceding AI, the lowest P4c preceding AI (P4c pre-AI) that followed a P4c-decline, and the interval between P4c-decline and AI were evaluated, as well as the interval between AI and onset of luteal phase, and P4c at early diestrus (4.5 ± 0.6 d post-AI), mid diestrus (10.0 ± 0.6 d post-AI), and late diestrus (14.1 ± 0.6 d post-AI). Data were analyzed using logistic regressions, and comparisons made based on quartiles and cut-points established by receiver operating characteristic curve analysis. Overall probability of pregnancy was 32.0%. Parameters associated with reduced probability of pregnancy (represented as percentage points decrease in the probability of pregnancy) were (1) luteal phase length >14.4 d (7.6% decrease), (2) P4c peak ≤24.7 ng/mL (4.5% decrease), (3) P4c pre-AI >0.5 ng/mL (5.5% decrease), (4) interval between P4c-decline and AI of >1.6 d (4.0% decrease), (5) interval between AI and onset of luteal phase of <7 or >11 d (9.3 and 12.1% decrease, respectively), and (6) P4c at early diestrus ≤0.7 or >3.5 ng/mL (15.2 and 6.7% decrease, respectively), (7) P4c at mid diestrus ≤12.4 ng/mL (12.5% decrease), and (8) P4c at late diestrus ≤22.7 ng/mL (9.7% decrease). The parameters of luteal activity associated with reduced probability of pregnancy established here could be used as benchmarks while developing recommendations to improve reproductive performance in herds using in-line milk progesterone monitoring.