The paper derives heteroscedasticity tests for the least weighted squares, which is a robust regression method with a high breakdown point. The methods are illustrated on numerical examples with real data.