An approach to studying reciprocal-effects in a cross-sectional data is to apply nonrecursive structural equation models. This article takes a theoretical approach to study the identifiability problem of reciprocal-effect models. Identifiability conditions are presented under a variety of settings.