Programming microworlds are being used for introducing students to programming for many years. Although many professors and school teachers report positive results from using programming microwords, these results are usually based on anecdotal evidence rather than rigorous empirical evaluation. A question that has not been answered yet with certainty is whether the knowledge acquired in the context of a programming microworld is transferred to a conventional programming language. In an attempt to investigate this issue we used a specially designed middle-term exam and a questionnaire at the end of an undergraduate Object-Oriented Programming (OOP) course. The course uses the programming microworld objectKarel for a short introduction to the most fundamental OOP concepts, and then the environment BlueJ and Java for presenting the real thing. The analysis of students' replies in the questionnaire shows that the introduction to OOP with the microworld helps the vast majority of students to comprehend fundamental concepts of OOP and what is more important is that this knowledge is transferred to Java afterwards. These results are supported by the comparative analysis of students' performance in written exams that took place in the context of two distinct offerings of the course, prior and after the use of the microworld in it. (Contains 3 figures and 4 tables.)