Is the size of a semantic-priming effect a function of the strength of the semantic context? This issue was examined in four studies using a single categorically related prime as the weaker context and four categorically related primes as the stronger context. Results indicate that, independently of prime-target SOA, four primes provide a larger priming effect than a single prime in a lexical-decision task, but not in a naming task. These data provide further support for the argument that different mechanisms mediate priming in the two tasks. In particular, only the lexical-decision task appears to be susceptible to higher-level processes that can be influenced by the number of primes. Priming in a naming task appears to be driven by more automatic processes. Possible accounts of this multiple-prime advantage in lexical-decision tasks are considered.