Coastal regions exhibit strong geographic patterns of nonnative species richness. Most invasions in marine ecosystems are known from bays and estuaries, where ship-mediated transfers (on hulls or in ballasted materials) have been a dominant vector of species introductions. Conspicuous spatial differences in nonnative species richness exist among bays, but the quantitative relationship between invasion magnitude and shipping activity across sites is largely unexplored. Using data on marine invasions (for invertebrates and algae) and commercial shipping across 16 large bays in the United States, we estimated (1) geographic variation in nonnative species richness attributed to ships, controlling for effects of salinity and other vectors, (2) changes through time in geographic variation of these ship-mediated invasions, and (3) effects of commercial ship traffic and ballast water discharge magnitude on nonnative species richness. For all nonnative species together (regardless of vector, salinity, or time period), species richness differed among U.S. coasts, being significantly greater for Pacific Coast bays than Atlantic or Gulf Coast bays. This difference also existed when considering only species attributed to shipping (or ballast water), controlling for time and salinity. Variation in nonnative species richness among Pacific Coast bays was strongly affected by these same criteria. San Francisco Bay, California, had over 200 documented nonnative species, more than twice that reported for other bays, but many species were associated with other (non-shipping) vectors or the extensive low-salinity habitats (unavailable in some bays). When considering only ship- or ballast-mediated introductions in high-salinity waters, the rate of newly detected invasions in San Francisco Bay has converged increasingly through time on that for other Pacific Coast bays, appearing no different since 1982. Considering all 16 bays together, there was no relationship between either (1) number of ship arrivals (from foreign ports) and number of introductions attributed to ships since 1982 or (2) volume of foreign ballast water discharge and number of species attributed to ballast water since 1982. These shipping measures are likely poor proxies for propagule supply, although they are sometimes used as such, highlighting a fundamental gap in data needed to evaluate invasion dynamics and management strategies.