In organisms with complex life cycles, fitness often increases with body size at the transition from larva to adult. The translation of larval size into fitness, however, can depend on the source of size variation, with size, per se, not always increasing adult success. In parasitic worms, many factors influence larval growth, but little is known about the consequences of this growth variation. We examined how the size of the tapeworm Schistocephalus solidus in its copepod first intermediate host affects infection success and growth in the stickleback second host. Moreover, we assessed whether the conspicuous growth variation caused by copepod size is fitness-relevant. Using larvae of the same age, we found that larger worms had a substantially higher infection probability and they tended to still be slightly larger after several months of growth in fish. However, big larvae from bigger copepods did not have higher fitness, suggesting that being large relative to the host, but not necessarily large in general, is important. These findings clarify some aspects of the life history strategy of S. solidus (e.g. why there is a flat ontogenetic reaction norm across copepod stages), but also raise questions (e.g. why growth costs have been hard to document). More generally, our results indicate that larval size can correlate with fitness in helminths, but that not all size variation is predictive of success in the next host.