We compared detectability of a dotted line masked by random-dot noise for the amblyopic versus non-amblyopic eye of two strabismic amblyopes. Small but consistent deficits in the amblyopic eye of these observers were found, and shown to be limited to dotted-line targets composed of greater than seven dots (with performance being normal for targets of less than seven dots). These deficits were unrelated to impaired visual acuity, impaired sensitivity to dot density, and differential positional uncertainty between the eyes of our observers. The deficits were also unlikely to be due to CSF losses due to abnormal low-spatial-frequency filters involved in detecting long chains of collinear dots. Instead, the results of simulations indicate that the inefficiency in utilising large numbers of dots is due to deficits of global, integrative processes in strabismic amblyopes. These simulations also show that while neither undersampling nor positional uncertainty of inputs into integrative processes can themselves account for the amblyopic deficits, if such abnormal inputs lead to the development of stunted integrative processes then impaired sensitivity to long chains of collinear dots is indeed predicted.