A collaborative analysis was undertaken of 226 karyotypes with structural chromosome abnormalities diagnosed primarily with low level banding resolution, about 300 to 400 bands per karyotype. We estimate that in this series, use of low level banding was required to detect about 78% of pericentric inversions, about 51% of reciprocal translocations, about 47% of all balanced translocations, about 35% of unbalanced rearrangements other than rings, Robertsonian translocations and extra structurally abnormal chromosomes, about 11% of all unbalanced rearrangements, and about 35% of all structural abnormalities. Adjustment factors derived from these figures were applied to prevalence and mutation rates of structural mutation rates derived from published large scale studies of livebirths. Had low level banding been used in these earlier studies we estimate that the rate of all structural abnormalities would have been about 60% higher than those reported (3.8 per 1000 vs. 2.3 per 1000 in the original studies). The increase is much higher for balanced abnormalities, 75% (3.4 per 1000 vs. 1.9 per 1000), than for unbalanced abnormalities, 5% (0.42 per 1000 vs. 0.405 per 1000). The increase in mutation rates for de novo cytogenetic abnormalities was similarly, considerably higher after such adjustment: the rates per 100,000 gametes increased from 18.0 to 35.0 for balanced rearrangements, from 8.2 to 10.1 for unbalanced abnormalities and from 26.2 to 45.1 for all abnormalities. These estimates illustrate the difference even low level banding makes to detection of structural cytogenetic abnormalities and why contemporary studies using such methods cannot be compared with earlier large scale population studies or livebirths without some type of adjustment such as those suggested here.(ABSTRACT TRUNCATED AT 250 WORDS)