In industrialized countries the share of elderly subjects is increasing. Hence, diseases or symptoms associated with aging are more common than they were in the past. As a consequence, more effort is invested into research analyzing the effects of aging on the motion and cognition. However, economical and flexible methods to measure motion and its cross-effects with cognition are still missing. Therefore, we developed a new approach which neither requires a specific location, large infrastructural requirements, nor does it require large investments. We base our setting on match-box sized inertial measurement units (IMUs) attached to the participants’ legs. 47 elderly subjects participated in our study where we analyzed the interplay between cognitive load and gait features. We show that it is feasible to automatically detect episodes of interest, e.g. straight path, during walking periods of a subject only using IMU data. Our approach detects the steps autonomously and calculates gait features without supervision. The results demonstrate that cognitive load induces a significant increase (p = 0.007) in step-duration variability from 16ms (baseline) to 21ms (load). Our findings demonstrate that IMUs are a proved alternative to static setups that usually require a non-trivial infrastructure, e.g. optical movement tracking.