Inference-making is a central element of successful reading comprehension, yet provides a challenge for beginning readers. Text decoding takes up cognitive resources which prevents beginning readers from successful inference-making and compromises reading comprehension. Listening does not require any decoding and could therefore offer a less demanding context to practice inference-making. The present study examined whether stimulating inference-making in a listening context is more effective and less cognitively demanding for beginning readers than a reading context. In three experiments, Dutch second grade children read two narratives and listened to two narratives. Inference-making was stimulated by asking them inferential questions during reading or listening and we compared this to a no-questioning control condition. After each narrative, we measured cognitive load and comprehension. It was expected that inferential questioning would increase cognitive load and negatively affect reading comprehension, but positively affect listening comprehension. The results indeed showed that inferential questioning increased cognitive load, but did not lead to differences in performance on open-ended comprehension questions (Experiment 1 & 2). When measuring comprehension with a free recall protocol (Experiment 3), we found a negative effect on total recall in both the reading and listening conditions. Taken together, we found no support for the hypothesized interaction. This raises questions about the effectiveness of inferential questioning for reading and listening comprehension of beginning readers, and whether listening is a good modality for improving inference-making.