The dissertation deals with the relationship between job insecurity and mental health from a spillover-crossover perspective. Intra-individual transmission of job insecurity from one life domain of an individual to another life domain of an individual as well as interindividual transmission of job insecurity across individuals are examined based on theoretical considerations of the Spillover-Crossover Model by Bakker and Demerouti (2013). Longitudinal dyadic data of heterosexual couples living together in one household in Germany between 2002 and 2012 is thereby used to analyze spillover and crossover of job insecurity to mental health as well as different vulnerability of job insecurity for women and men theoretically motivated by the Social Role Theory (Eagly and Wood, 2011). Based on multilevel modeling of longitudinal data and multilevel modeling of dyadic data, a three-step estimation strategy for multilevel modeling of longitudinal dyadic data of job insecurity and mental health within an Over-time Standard Actor-Partner Interdependence Model is developed, estimated and evaluated. The results confirm the intra-individual and the inter-individual transmission of job insecurity. Individuals who suffer from job insecurity on average display a worse mental health status than individuals who do not suffer from job insecurity. In addition, individuals who are in a relationship with partners who suffer from job insecurity on average display a worse mental health status than individuals being in a relationship with partners who do not suffer from job insecurity. A different vulnerability of job insecurity for women and men to the disadvantage of male individuals can be found as well. The results are proofed by including control variables, controlling for temporal asymmetry and considering unobserved heterogeneity and omitted variables. In conclusion, the empirical findings display a wider and more comprehensive consequence of job insecurity than it is identified in previous research until now. Furthermore, a contribution to the methodical approach for longitudinal dyadic data can be made.