Quality variations in manufacturing are the root causes for product high infant failure rate. A structural model of quality variations in manufacturing composed of part-level, component-level and system-level is put forward firstly. Then the bi-variant Gumbel Copula for positive correlation quantification is tailored to obtain the correlation dependence of the three levelled variations, and the impact of the structural but correlated quality variations on product infant failure rate is conducted systematically under Weibull distribution. Finally, a case study is carried out to verify the proposed method.