Cognitive load theory (CLT) predicts that bimodal processing of instructional material decreases the level of extraneous cognitive load, whereas increased training variability increases the level of germane cognitive load. It was hypothesized that a combination of these strategies leads to optimal learning, especially in older adults. Forty young and forty elderly learners were trained to solve complex problems. The results showed that bimodal training leads to lower cognitive load than unimodal training. Furthermore, random presentation of examples (high variability) led to higher performance than blocked presentation (low variability) in both age groups. However, there was no combined effect of modality and variability. Moreover, the elderly did not take disproportionate advantage of the bimodal and random conditions. It was concluded that these training methods bear important potential in the light of lifelong learning.