There are a variety of spline formulations for specifying smooth curves.In theory, many of these are equivalent in terms of the families of curves that can be obtained.In practice, each formulation encourages the implementation of certain styles of interaction in a curve editor and discourages other styles.This paper presents the design, implementation, and analysis of four experiments to illustrate a general shape-matching paradigm for studying interactive curve manipulation.We introduce the new shape-matching paradigm as a methodology for analyzing user performance.We then report on a series of studies that employ this paradigm to explore the relationship between a commonly used interaction style for curve design and various spline formulations, Controlled experiments were performed to test the hypothesis that the style of interaction implicit in the movement of single design points will influence the ease with which curve manipulation can be performed using a variety of spline formulations.Although such interactions are of a simple form, the experience gained in these studies has already enabled us to undertake more extensive experiments with more sophisticated interaction styles.The specific experiments reported here compare five formulations of cubic splines (B-splines, B6zier splines, Catmull-Rom splines, and two versions of interpolating splines with C 2 continuity), Data from the experiments show that, for the particular shape-matching tasks investigated, there is a significant performance difference among the five formulations.The B-spline formulation was best for the task, both in terms of match quality and in terms of the time required to Financial support for this research was provided by the