The objective was to characterize activity before artificial insemination (AI) in cows that did or did not conceive in grazing and non-grazing conditions. Holstein cows (n = 310) from a herd in Northern Colorado (2,800 milking cows) were affixed with a pedometer (IceQube IceRobotics, Edinburgh, UK) on one leg at 12 ± 8 DIM to monitor steps (ST; n/d), lying time (LT; min/d), and lying bouts (LB; n/d) for 7 mo. Milk yield (MY; kg/d) and reproductive data were obtained from on-farm software. Reproduction was managed through heat detection by use of tail head painting and AI technicians were unaware of activity data. Cows were maintained in free stall barns and had access to pasture during the final 58 d of the study. Activity parameters were analyzed as continuous and categorical variables (high vs. low relative to the median). Generalized linear models and logistic regression analysis were used to evaluate activity on the day before AI (d-1) and subsequent pregnancy. Models included DIM, parity, THI, technician, and sire as covariables. Overall, 67% of the cows were pregnant at the end of the study. There were 551 and 221 AI during the non-grazing and the grazing period (27.6% vs. 30.8% conception). A significant effect on ST was determined for the interaction between grazing and AI outcome (P = 0.02), as conception was associated with greater or lower ST, depending on access to grazing. Subsequent analyses by grazing category indicated no differences in ST for cows that conceived or remained open. LT and LB were not affected by access to grazing. LT at d-1 was greater in cows that conceived (537 ± 16 vs. 498 ± 9; P < 0.05). Similarly, LB at d-1 was greater in cows that conceived (9.26 ± 0.33 vs. 8.5 ± 0.21; P = 0.05). Daily MY was not associated with AI outcome. The logistic regression analyses considering activity variables divided in categories indicated that the only parameter partially associated with conception was ST: During the non-grazing season, the odds (95% CI) of pregnancy in cows in the high ST category were 2 (1.2–3.33) times the odds of cows in the low ST category. The level of variation on activity behavior before AI was partially different for cows that conceived or failed to conceive in grazing and non-grazing conditions.