Two major families of transcription factors (TFs), basic helix-loop-helix (bHLH) and homeodomain (HD), are known to be involved in cell fate identity. Some recent findings suggest that these TFs are used combinatorially to code for cellular determination in the retina. However, neither the extent nor the efficiency of such a combinatorial coding mechanism has been tested. To look systematically for interactions between these two TF types that would address these questions, we used a matrix analysis. We co-expressed each of six retinally expressed bHLH TFs (XNeuroD; XNgnr-1; Xath3; Xath5; Xash1; Xash3) with each of eight retinally expressed HD TFs (XRx1; XOptx2; XSix3; XPax6; XOtx2; XOtx5b; XBH; XChx10) in retinal progenitors of Xenopus laevis using targeted lipofection. The effects of each of these combinations were assayed on the six major cell types in the retina: Retinal ganglion cells (GCs), Amacrines (ACs), Bipolars (BCs), Horizontals (HCs), Photoreceptors (PRs), and Muller cells (MCs), creating 288 result categories. Multiple-way ANOVA indicated that in 14 categories, there were interactions between the two TFs that produced significantly more or less of a particular cell type than either of the components alone. However, even the most effective combinations were incapable of generating more than 65% of any particular cell type. We therefore used the same techniques to misexpress selected combinations of three TFs in retinal progenitors, but found no further enhancements of particular cell fates, indicating that other factors are probably involved in cell type specification. To test whether particular combinations were essential for horizontal fates, we made VP16 and EnR fusion constructs of some of the factors to provide dominant negative transcriptional activities. Our results confirmed that normal activities of certain combinations were sufficient, and that individually these activities were important for this fate.