This study examines elicited imitation (EI) both as a measure of implicit grammatical knowledge and more global semantic and syntactic knowledge. It also examines whether length affects the difficulty of EI tests when they contain both grammatical and ungrammatical items. Fifty language learners took an EI test and an oral narrative task. The data were analyzed once for the accuracy of a single target structure, third person "-s", and then for the global semantic and syntactic accuracy. Moderate correlations were recorded between the learners' third person accuracy scores on the oral narrative task and the grammatical and ungrammatical items of the EI test (r?=?0.62 and r?=?0.66 respectively). A moderate correlation was also found between the learners' global semantic and syntactic scores on the EI test and the oral narrative task (r?=?0.51). Furthermore, unlike the grammatical items, no significant correlation was recorded between length and the performance of the learners on the ungrammatical items. The findings underscore the need to validate EI against other tests of productive skills especially when EI is used as a measure of global language proficiency. Suggestion is made to explore structural-specific factors that contribute to difficulty level of EI tests.