The molecular genetic diagnosis of spinal muscular atrophy (SMA) has recently been complicated by the identification of two candidate genes, which are often deleted in affected individuals but are also occasionally deleted in apparently unaffected carriers. We present a compilation of genotypes, from our laboratory and recent reports, for the survival motor neuron (SMN) and neuronal apoptosis inhibitor protein (NAIP) genes. Bayesian analyses were used to generate probabilities for SMA when deletions are present or absent in SMN. We found that when the SMN(T) exon 7 is deleted, the probability of SMA can reach greater than 98% in some populations, and when SMN(T) is present, the probability of SMA is approximately 17 times less than the prior population risk. Deletion of NAIP exon 5, as well as SMN(T) exon 7, is associated with a 5-fold increased risk of type I SMA. Case studies are used to illustrate differing disease risks for pre- and postnatal testing, depending on the presence of information about clinical status or molecular results. These analyses demonstrate that deletion screening of candidate genes can be a powerful tool in the diagnosis of SMA.