A method is proposed for the analysis of clustered, continuous data arising from matched familial aggregation studies. The problem is suggested by a study designed to estimate the familial aggregation of obstructive sleep apnea (OSA). The study design uses ‘proband’ sampling, in which individuals identified at sleep clinics as having a high degree of OSA (‘cases’) are matched to neighborhood ‘controls’, individuals with normal levels of OSA. The resulting data are used to develop estimates of familial correlations for indices of hypopnea/apnea derived from the sleep monitoring of individual family members. The properties of conditional multivariate normal distributions and random-effects models are used to accommodate proband sampling and matching effects. Examples and simulations are used to demonstrate the importance of applying the proposed procedures.