We investigated 33 individuals (21 carriers) from one family with a pericentric inversion involving a large part of chromosome 1 (1p36.1----1q32). In addition, we investigated 15 individuals (10 carriers) from another family with a paracentric inversion of a small part of chromosome 1 (1p32----1p36.1). In each family, the index patient was ascertained because three miscarriages had occurred. Each carrier of these inversions was phenotypically normal. If the miscarriages of the index patients are excluded, the frequency of recognized miscarriages among the carriers of childbearing age was 9% (4 of 46) for the family with pericentric inversion and 17% (4 of 23) for the family with paracentric inversion. One of the pericentric inv(1) carriers had had a stillborn daughter. The carriers of the pericentric inversion who were of childbearing age had 41 children; carriers of the paracentric inversion who were of childbearing age had 19 children. No live-born children with birth defects were observed in either family. This evidence, together with the low frequency of miscarriages, suggests that crossover within the inversion loop occurs much less frequently than might be expected from the large size of this inversion. Our investigation suggests that the risk of recognized miscarriages, stilbirths, and live-born children with recombinant chromosomes who have birth defects may be much lower for inv(1) carriers than previously reported. The risk of having a malformed child because of a recombinant chromosome is probably less than 3% for carriers of the pericentric inversion and less than 6% for the carriers of the paracentric inversion.