Abstract: To fully expolit the advantage of precise GPS carrier phase measurements the integer ambiguities in the observations must be resolved.There are several ambiguity resolution techniques developed in the last decade.A common criterion used in these techniques to select the best set of ambiguities is the minimum quadratic form of the residuals.Because the observations contain various types of errors,the resolved integer ambiguities are with some degree of uncertainty.To obtain a reliable result the selected ambiguities must be tested against the other possible ones,which is called the validation of resolved ambiguities.There exist several approaches,each of which,however,has some limitation and shortcomings.This research has developed a methodology based on the concept of model discrimination.Given type Ⅰ and type Ⅱ errors two boundary values are derived.The solution with the minimum variance factor is validated using these two values.Compared with the existing approaches,the proposed one is theoretically sound,and the decision made based on it will have a clear indication of confidence in accepting the resolved ambiguities and rejecting the others.