To analyze variance in a triadic variable, Bond, Horn, and Kenny (1997) have proposed a Triadic Relations Model. Here we extend this model to analyze the covariances between triadic variables. A bivariate version of the Triadic Relations Model is specified, and estimation methods are presented. These can be used to decompose the covariance between two triadic variables into thirty-three covariance components. Interpretations and an example of this analysis are offered. Applications of this model and alternative techniques are noted.