Simple but objective criteria for evaluating the quality of DTMs, derived from digital contour lines, are defined. Such criteria are then applied to terrain models, obtained by means of four different procedures, which refer to three sample areas characterized by complex morphological settings. The results of the comparison shows that each DTM is affected by one or more types of error or pitfall; however, one TIN generator and one grid interpolator proved to be capable of effectively producing terrain models which largely reflect the ground morphology as expressed by the input contour lines.