AbstractVisuospatial inattention is one of the deficits which can have a negative effect on activities in daily living performance in stroke patients. Four cancellation tests were administered to 43 stroke patients (34 right lesion and 9 left lesion) to determine the reliability and validity in measuring visuospatial inattention deficit. For the total group, test–re‐test reliabilities on the Albert Test, the Schenkenberg Line Bisection Test (SLBT) cancellation, the Structured Chinese Word Cancellation Test (SCWCT) and the Random Chinese Word Cancellation Test (RCWCT) ranged from 0.79 to 0.96. Among the four cancellation tests, the RCWCT had the highest reliability (0.96). Reliability coefficients were also calculated for the subgroup with right brain lesion with similar results. High intercorrelation coefficients showed that the four tests evaluate the same construct at a significant level. The correlation (r) was used to estimate the size of the effect of the linear relationship between lesion side and omission scores of the tests. The results showed moderate‐to‐large effect sizes of the side of lesion on the performance and indicated that the RCWCT, the SCWCT, the SLBT and the Albert Test could all be used to discriminate right brain lesion from the left. In addition to showing excellent reliability and construct validity, the RCWCT was the most difficult of the tests, and so the most likely to identify inattention in Chinese indi