It is often assumed that in situ target strength (TS) measurements from dispersed fish are representative of the surveyed schooling fish. For in situ TS measurements of orange roughy in deep water, it has been difficult to validate the target species, individual lengths, and tilt angles and how representative these are of schooling fish. These problems have been addressed by attaching an acoustic optical system (AOS) to a trawl net. The AOS enables in situ measurements of TS and volume backscattering strength (Sv) at 38 and 120 kHz with optical verification of species and stereo camera measurements of fish length and tilt angle. TS estimates believed representative of the schooling population were derived by (1) weighting the frequency-dependent TS values by the Sv frequency difference distribution of orange roughy schools and (2) weighting the in situ TS measurements with an assumed tilt angle distribution. The 120-kHz TS estimates were less sensitive to variations in frequency difference and tilt angle, suggesting that this frequency may be better for biomass estimates than 38 kHz, the traditional survey frequency. Computations performed with an anatomically detailed scattering model agree with measurements of TS at both frequencies over a range of tilt angles.