Satchmo (edgar and sjolander, 2003) makes different predictions of alignable regions in different subgroups. Experiments were conducted that provide a proof-of-concept for developing an iterated form of the satchmo alignment method. Whereas the previously existing algorithm terminates after the progressive stage, we now iteratively refine the alignment in a second stage. The new algorithm was tested against the prefab benchmark alignment dataset of structurally aligned proteins to obtain a score for the predicted (satchmo) alignments. Our limited trials show that significant improvement can be obtained for some datasets. We seek to implement a second track for predicted secondary structure to the profiles for scoring and alignment. This should enable us to improve the accuracy of tree topology and alignment of two profiles when the sequences included in each profile are very distantly related.