Butler (2003) used first arrival dates (FADs) of 103 migrant birds in northeastern USA and found that both long-distance migrants (LDMs; wintering south of the USA) and short-distance migrants (SDMs; wintering in the southern USA) arrived earlier in the second half of the 20th century than they had in the first, consistent with scenarios of global warming; the trend was stronger in SDMs. Using FADs to characterize migration systems can be problematic because they are data from one tail of a distribution, they comprise a mostly male population and they may not correlate well with the balance of the migration period. FADs also provide no information about autumn migration. This paper uses a banding dataset from Long Point Bird Observatory, Ontario, for 14 passerines for a period of global warming (1975–2000), taking these issues into account. The data were filtered to minimize effects of unequal netting effort (147 491 resulting records), and the passage dates then calculated in each season of each year for the 1st, 2nd and 3rd quartiles for regression analysis. Only two of 13 species analysed in the spring showed significantly earlier passage times, although the overall trend was towards earlier spring migration, especially among SDMs. Autumn responses were more prevalent, however, and in some cases more dramatic with six of 13 species showing delayed migration (four SDMs, two LDMs). Two LDMs exhibited earlier autumn migration. Where earlier spring migration occurred, both sexes appeared to contribute to the change. Where delayed migration occurred in autumn, both sexes and both adults and hatch-year birds appeared to contribute in at least some cases. The spring FAD results are consistent with those of Butler, but when the whole migration is considered, change is far from universal in spring and is in fact more substantial and complex in autumn.