A printer inverse map is usually represented as a multivariable lookup table, associating points in the printer's output color space with points in the printer's input color space. This lookup table is an essential component in many print quality enhancement algorithms. It is often desirable to have a printer inverse table with input nodes regularly spaced on a sequential plane. Thus, the computation of this lookup table requires the interpolation of irregularly sampled multidimensional data, coming from experiments used to determine the printer forward map. Existing computational techniques do not provide an accurate printer inverse map from irregularly sampled data. In this paper we introduce a new Iteratively Clustered Interpolation (ICI) algorithm to compute an accurate inverse table from irregularly sampled color data. This algorithm is based on a gradient optimization method with initial points generated through a novel iterative technique. Experimental results are included to show the effectiveness of this algorithm in comparison with other techniques.