Background. Familial factors, which are partly genetic, influence risk for phobias. Prior family and twin studies, however, were based on a single lifetime assessment, which may be only moderately reliable.

Methods. We obtained, 8 years apart, two assessments of lifetime history of five unreasonable fears and phobias (agoraphobia and social, situational, animal and blood-injury phobia) from face-to-face and telephone interviews from 1708 individual female twins from a population-based registry. We also obtained, 1 month apart, test–retest reliability on 192 twins. We fitted, using the program Mx, a measurement model that estimates the role of genetic and environmental risk factors correcting for measurement error.

Results. Short-term reliability of the five phobias was modest (mean κ=0·46), but higher than long-term stability (mean κ=0·30). Unreliability occurred both for subject recall of unreasonable fears and for interviewer assessment of which fears constituted phobias. Examining fears and phobias together, in a multiple threshold model, results suggested that twin resemblance was due solely to genetic factors, with estimated total heritabilities, corrected for unreliability, of: any 43%, agoraphobia 67%, animal 47%, blood/injury 59%, situational 46% and social 51%. With the exception of animal phobia, similar results were obtained analysing phobias alone.

Conclusions. Lifetime histories of unreasonable fears and phobias assessed at personal interview have substantial unreliability. Correcting for unreliability, the liability to fears and their associated phobias is moderately heritable. Individual-specific environmental experiences play an important role in the development of phobias, while familial–environmental factors appear to be of little aetiological significance.