Abstract Macroscopic maturity data collected between January and March of 1978–2004 during groundfish surveys were analysed to determine the frequency of skipped spawning for Atlantic cod (Gadus morhua) populations in the Northwest Atlantic. Nearly 18% of the 6583 potential female spawners sampled during this period would not have spawned in the subsequent spawning season. The frequency of skipped spawning was much higher on the Flemish Cap (NAFO Division 3M, ∼33%) and off the south coast of Newfoundland (Subdivision 3Ps, ∼21%) than that in three areas off eastern Newfoundland and southern Labrador (Divisions 3L, ∼8%; 3K, ∼3%; and 2J, <1%). Annual variation in the frequency of skipped spawning was apparent in all areas. Skipped spawning was most common in fish of medium size (60–79 cm). Non-reproductive cod were characterized by lower liver condition than reproductive cod, but not a lower gutted condition. Generalized linear models suggested that relative liver weight was the best predictor of spawning probability, and that age, length, and relative body condition were not good predictors. Models of spawning probability based on relative liver condition suggest that female fish in poor condition in 3KL are more likely to spawn than fish of equal condition in areas 3M and 3Ps.