There is evidence that prediction takes place during simultaneous interpreting. But how does such prediction take place? We report the results of an eye-tracking study that investigated whether prediction in 32 L1-French L2-English participants who performed a simultaneous interpreting task involves two stages. Participants heard sentences produced by either a male speaker (who mentioned stereotypically masculine objects) or a female speaker (who mentioned stereotypically feminine objects). On hearing a constraining verb (e.g. wear), participants looked predictively at an object that was stereotypically masculine (a tie) and one that was stereotypically feminine (a dress). But they then focused their looks on the object that stereotypically agreed with the speaker’s gender, in a manner broadly similar to prediction during language comprehension. We conclude that prediction in simultaneous interpreting from L2 into L1 involves two stages, and that interpreters integrate speaker perspective during the second stage.