Sonbul and Schmitt (2013) showed that exposure to second language (L2) collocations in reading texts can produce gains in explicit knowledge, but they found no evidence of gains in implicit knowledge. This study is a conceptual replication and extension of Sonbul and Schmitt's research. Sixty-two advanced English as a second language (ESL) speakers read texts containing repeated occurrences of low-frequency medical collocations in three sessions over 2 days. The incidental learning treatments included reading only (no typographic enhancement), bolding, and bolding-plus-glossing. Collocational knowledge was assessed in tests of explicit knowledge (cued recall, form recognition) and a test of implicit knowledge (primed lexical decision). Repeated exposure to bolded collocations produced greater explicit knowledge than repeated exposure to typographically unenhanced collocations. Evidence of implicit knowledge development was observed in the unenhanced (reading only) treatment but not in the remaining treatments. These results replicated Sonbul and Schmitt's findings for explicit knowledge and extended their findings for implicit knowledge.