Duplicate coverage involves those individuals who hold compulsory health insurance with the public sector and have additional coverage with the private sector. The additional insurance covers costs for outpatient and inpatient care, income loss and hospital daily allowances. The number of persons who took out additional coverage has been steadily increased. This increase can be linked to two main factors: the shortage in the benefits package and the introduction of the reform act (January the 1st 2004). Basically, members of the public insurance sector have to make co-payment of 10 percent for all health care services and drug prescription (maximum 2 percent of the annual pre-tax income). Costs of transportation and dental prosthesis have been also excluded from the benefits package. It uses the SOEP German database for estimate an demand model for health services, given the simultaneity of the choices to take duplicate coverage and the level of health services (measured like number of visits), we estimate a negative binomial model to measure the impact of the duplicate coverage on the health service demand, we also estimate a a Full Information Maximun Loglikelihood (FIML) known in this case as an Endogenous Switching Poisson Count Model and we compare this results with the standard maximum log likelihood (ML) estimators of the negative binomial model. The Results show that there is a positive difference on the level of health services demanded when there is a duplicate coverage. We found also that there is evidence to think that in Germany there is a feedback between duplicate coverage and the demand of health services.