Friedreich ataxia (FRDA) is associated with a GAA-trinucleotide-repeat expansion in the first intron of the FXN gene (9q13-21), which encodes a 210-amino-acid protein named frataxin. More than 95% of patients are homozygous for 90-1,300 repeat expansion on both alleles. The remaining patients have been shown to be compound heterozygous for a GAA expansion on one allele and a micromutation on the other. The reduction of both frataxin messenger RNA (mRNA) and protein was found to be proportional to the size of the smaller GAA repeat allele. We report a clinical and molecular study of 12 families in which classical FRDA patients were heterozygous for a GAA expansion on one allele. Sequence analysis of the FXN gene allowed the identification of the second disease-causing mutation in each heterozygous patient, which makes this the second largest series of FRDA compound heterozygotes reported thus far. We have identified seven mutations, four of which are novel. Five patients carried missense mutations, whereas eight patients carried null (frameshift or nonsense) mutations. Quantitation of frataxin levels in lymphoblastoid cell lines derived from six compound heterozygous patients showed a statistically significant correlation of residual protein levels with the age at onset (r = 0.82, p < 0.05) or the GAA expansion (r = -0.76, p < 0.1). In the group of patients heterozygous for a null allele, a strong (r = -0.94, p < 0.01) correlation was observed between the size of GAA expansion and the age at onset, thus lending support to the hypothesis that the residual function of frataxin in patients' cells derive exclusively from the expanded allele.