The Symbol Digit Modalities Test (SDMT) and Digit Vigilance Test (DVT), both well-recommended attention tests for schizophrenia, are measures of switching and sustained attention, respectively. The purpose of this study was to investigate the test-retest reliability of the two attention tests in schizophrenia. A rater administered both tests on 147 participants with schizophrenia twice at a 1-week interval. Test-retest reliability was determined through the calculation of the intra-class correlation (ICC) coefficient. We also carried out the Bland-Altman analysis, which include a scatter plot of the differences between test and retest against their mean. System biases were evaluated by use of a paired t-test. The ICC for the SDMT was 0.87 and that for the DVT was 0.83. The limits of agreement (LOAs) of the SDMT and DVT were 11.5 to -9.9 correct responses and 156.3 to -249.2 s, respectively. The mean difference scores of the SDMT and DVT were 1.5 (4.7% of the first session mean; p= .002) and -46.4 (7.6% of the first session mean; p< .001). The ICCs show that the SDMT and DVT are stable measures across assessment in different sessions in schizophrenia. However, the paired t-test indicates a practice effect, and the LOAs show large variations. Thus, the SDMT and DVT are reliable for a group of subjects but limited for individual subjects with schizophrenia in 1-week interval clinical trials.