Mandarin Chinese, unlike Polish, has no voiced stops and it contrasts voiceless unaspirated and voiceless aspirated stops, whereas Polish contrasts voiced and voiceless unaspirated stops. A study on the perception of voicing contrasts in Mandarin showed that Polish advanced speakers of Mandarin consistently divide stops into lenis (/p t k/) and fortis (/ph th kh/). This paper is to show whether they are able to articulate these sounds in a native-like manner. A group of university students participated in our pilot study. They were all advanced learners of Mandarin (their L3). They were asked to imitate syllables they heard. Their performances were recorded and analysed by measuring VOT values of the target sounds. The results of the experiment reveal that long lag stops are generally produced with sufficient, native-like, aspiration, whereas short lag stops appear to be much more problematic and Polish advanced users of Mandarin, even though exposed to the model, do not always produce them correctly, i.e. they add voicing and articulate these sounds in a Polish-like manner.