Blepharophimosis syndrome (BPES), an autosomal dominant syndrome in which an eyelid malformation is associated (type I) or not (type II) with premature ovarian failure (POF), has recently been ascribed to mutations in FOXL2, a putative forkhead transcription factor gene. We previously reported 22 FOXL2 mutations and suggested a preliminary genotype-phenotype correlation. Here, we describe 21 new FOXL2 mutations (16 novel ones) through sequencing of open reading frame, 5' untranslated region, putative core promoter, and fluorescence in situ hybridization analysis. Our study shows the existence of two mutational hotspots: 30% of FOXL2 mutations lead to polyalanine (poly-Ala) expansions, and 13% are a novel out-of-frame duplication. In addition, this is the first study to demonstrate intra- and interfamilial phenotypic variability (both BPES types caused by the same mutation). Furthermore, the present study allows a revision of the current genotype-phenotype correlation, since we found exceptions to it. We assume that for predicted proteins with a truncation before the poly-Ala tract, the risk for development of POF is high. For mutations leading to a truncated or extended protein containing an intact forkhead and poly-Ala tract, no predictions are possible, since some of these mutations lead to both types of BPES, even within the same family. Poly-Ala expansions may lead to BPES type II. For missense mutations, no correlations can be made yet. Microdeletions are associated with mental retardation. We conclude that molecular testing may be carefully used as a predictor for POF risk in a limited number of mutations.