EFL learners in two countries participated in two parallel experiments testing whether retention of vocabulary acquired incidentally is contingent on amount of task‐induced involvement. Short‐ and long‐term retention of ten unfamiliar words was investigated in three learning tasks (reading comprehension, comprehension plus filling in target words, and composition‐writing with target words) with varying “involvement loads”—various combinations of need, search, and evaluation. Time‐on‐task, regarded as inherent to a task, differed among all three tasks. As predicted, amount of retention was related to amount of task‐induced involvement load: Retention was highest in the composition task, lower in reading plus fill‐in, and lowest in the reading. These results are discussed in light of the construct of task‐induced involvement.