Visual neglect is a relatively common deficit after brain damage, particularly strokes. Cancellation tests provide standard clinical measures of neglect severity and deficits in daily life. A recent single-case study introduced a new variation on standard cancellation. Instead of making a visible mark on each target found, the patient made invisible marks (recorded with carbon paper underneath, for later scoring). Such invisible cancellation was found to reveal more neglect than cancellation with visible marks. Here we test the generality of this. Twenty three successive cases with suspected neglect each performed cancellation with visible or invisible marks. Neglect of contralesional targets was more pronounced with invisible marks. Indeed, about half of the patients only showed neglect in this version. For cases showing more neglect with invisible marks, stronger neglect of contralesional targets correlated with more revisits to ipsilesional targets for making additional invisible marks upon them. These results indicate that cancellation with invisible marks can reveal more neglect than standard cancellation with visible marks, while still providing a practical bedside test. Our observations may be consistent with recent proposals that demands on spatial working memory (required to keep track of previously found items only when marked invisibly) can exacerbate spatial neglect.