This study investigates the cross-dialectal variability in the production of Mandarin alveolar and retroflex sibilants. A series of map tasks were designed to explore the effects of vowel context and prosodic prominence on the sibilant realizations in Beijing Mandarin and Taiwan Mandarin. Results show that the acoustic realizations of the alveolar-retroflex contrast varied greatly across different vowel contexts. In particular, the /u/ context was where the place contrast was the least distinct. Focal prominence was not found to affect the sibilant realizations in the spectral dimension, although the focus effect was manifested in syllable duration in that target syllables produced under focus condition were significantly longer than those in non-focus condition. While some speakers' data seem to be selectively subject to prosodic conditioning in certain vowel contexts, no overall contrast enhancement in response to focal prominence was observed, suggesting that prosodic prominence does not always lead to enhancement of phonologically contrastive features, especially place features.