Perceptual and response bias in estimating the proportion of the two segments of prebisected lines were disambiguated in a group of 121 patients suffering from left neglect by means of two variants of the Milner Landmark task (Milner et al., 1993). The first variant, LANDMARK-V, required a verbal response; the second variant, LANDMARK-M, required manual pointing. The paper reports and discusses the results obtained on each task and their correlations, as well as the relationships between either kind of bias and the intrahemispheric location of the lesion. It is argued that besides their usefulness as a diagnostic tool the proposed variants of the Milner Landmark task provide results that are worth further investigation in their own right.