A study investigated the problem of systematic oversights by students taking language passage correction (PC) tests. In these tests, students are asked to correct errors inserted in written prose passages. When errors are not overtly marked, students often overlook them. This study examined patterns of oversight in two PC tests used to evaluate skills in English as a second language (ESL). In one test, students had to consult a picture to correct semantic inaccuracies and other errors. In the other, students had to correct a written passage without consulting a picture. Results of the picture test were compared with scores on a standardized English test, results of the second test were compared with holistic evaluations of class essays, and for both tests, the results obtained from native and non-native English speakers were compared. Results of some of the analyses suggest that PC tests may be useful for research in second language learning and for second language instruction. However, some patterns in the test results suggest potential problems in interpretation. These include differences among native speakers' performances, discrepancies between PC and holistic evaluation results, and low inter-item correlations despite a high internal consistency measurement. (MSE)