Association between spatial gait parameters and adverse health outcomes in the elderly has not been sufficiently studied. The goal of this study is to evaluate whether the stride length or the step width predict falls, functional loss and mortality. We conducted a prospective cohort study on a probabilistic sample of 431 noninstitutionalized, older-than-64-years subjects living in Spain, who were followed-up for five years. In the baseline visit, spatial gait parameters were recorded along with several control variables, with special emphasis on known medical conditions, strength, balance and functional and cognitive capacities. In the follow-up calls, vital status, functional status and number of falls from last control were recorded. We found that a normalized-to-height stride length shorter than 0.52 predicted recurrent falls in the next 6 months with 93% sensitivity and 53% specificity (AUC: 0.72), and in the next 12 months with 81% sensitivity and 57% specificity (AUC: 0.67). A normalized stride length <0.5 predicted functional loss at 12 months with a sensitivity of 79.4% and specificity of 65.6% (AUC: 0.75). This predictive capacity remained independent after correcting for the rest of risk factors studied. Step-with was not clearly related to functional loss or falls. Both shorter normalized stride length (OR1.56; AUC: 0.62; p < 0.05) and larger step width (OR1.42; AUC: 0.62; p < 0.05) were associated with risk of death at 60 months; however, none of them remained as independent predictor of death, after correcting for other risk factors. In summary, spatial gait parameters may be risk markers for adverse outcomes in the elderly. Step length is independently associated with functional loss and falls at one year, after correction for numerous known risk factors.