The Prader-Willi syndrome usually occurs sporadically. About 60% of the patients show a chromosomal deletion of proximal 15q. There are only a few reports on familial recurrence. Two previous estimates give either a low but recognizable recurrence risk of 1.6% or a risk of less than 0.1%. The inconsistency is obviously due to differences in the stringency of the diagnostic criteria used and to different modes of ascertainment of cases. This gave rise to a review of the literature, leading to the following suggestions: (1) A deletion at 15q has not been found in a familial case of PWS, except in those cases where del(15q) is due to familial structural chromosome rearrangements. Hence, with de novo deletion, the recurrence risk should be nearly zero. (2) In cases with familial translocation, risk estimates depend on the variable nature of the familial chromosome mutation. (3) If only one child is affected and has no deletion at 15q, this may be a sporadic or an isolated familial case. For this situation I estimated an overall recurrence risk of 0.4%. (4) If two or more sibs are affected, I consider a risk of 50% that the next sib may also be affected. This estimate reflects the observed segregation ratio and is in accordance with the proposed genomic imprinting model in the etiology of PWS. Prenatal cytogenetic diagnosis in a pregnancy at risk for PWS is hampered since del(15q) is almost exclusively found in sporadic cases. Hence, this method is limited to those very rare cases where del(15q) is due to familial structural chromosome rearrangements.(ABSTRACT TRUNCATED AT 250 WORDS)