A controlled trial investigating the effect on conception of administration of 250 μg of gonadotropin-releasing hormone (GnRH) at artificial insemination (AI) in dairy cows in seasonal or split calving herds was conducted. Time of detection of estrus, body condition, extent of estrous expression, treatment, breed, age and milk production from the most recent herd test of the current lactation was recorded. Cows were tested for pregnancy with fetal aging between 35 and 135 days after AI. Sixteen herds provided 2344 spring-calved cows and 3007 inseminations. Logistic regression adjusting for clustering at herd level was used to examine the effect of treatment for first (2344) and second (579) inseminations separately. For first AI, treatment significantly improved conception rate in cows with milk protein concentrations of 3.75% or greater and for cows with milk protein concentrations between 3.00% and 3.50% and less than 40 days calved; increased conception rate from 41.2% to 53.4%. Treatment reduced conception rates in cows with milk protein concentrations of 2.75% or less. Treating only cows identified as responding positively to treatment (11% of all study cows) was estimated to increase first service conception rate in herds from 48.1% to 49.4%. There was no significant effect of treatment on conception to second AI, nor any significant interactions. These findings indicate that GnRH at AI should be limited to the sub-group cows most likely to respond. The positive effect of GnRH at AI may be mediated through improved oocyte maturation and/or improved luteal function, rather than by reducing AI-to-ovulation intervals.