This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues.Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. Abstract Objectives: To evaluate fall risk in stroke patients based on single-and dual-task gait analyses, and to investigate the difference between 2 cognitive tasks in the dual-task paradigm. Design: Prospective cohort study. Setting: Rehabilitation hospitals. Participants: Subacute stroke patients (NZ32), able to walk without physical/manual help with or without walking aids, while performing a verbal task. Interventions: Not applicable. Main Outcome Measures: Functional gait measures were Functional Ambulation Categories (FAC) and use of a walking aid. Gait measures were evaluated by an electronic walkway system under single-and dual-task (DT) conditions. For the single-task, subjects were instructed to walk at their usual speed. One of the DTs was a verbal fluency dual task, whereby subjects had to walk while simultaneously enumerating as many different animals as possible. For the other DT (counting dual task), participants had to walk while performing serial subtractions. After inclusion, participants kept a 6-month falls diary. Results: Eighteen (56.3%) of the 32 included patients fell. Ten (31.3%) were single fallers (SFs), and 8 (25%) were multiple fallers (MFs). Fallers (Fs) more frequently used a walking aid and more frequently needed an observatory person for walking safely (FAC score of 3) than nonfallers (NFs). Two gait decrement parameters in counting dual task could distinguish potential Fs from NFs: decrement in stride length percentage (PZ.043) and nonparetic step length percentage (PZ.047). Regarding the division in 3 groups (NFs, SFs, and MFs), only MFs had a significantly higher percentage of decrement for paretic step length (PZ.023) than SFs. Conclusions: Examining the decrement of spatial gait characteristics (stride length and paretic and nonparetic step length) during a DT addressing working memory can identify fall-prone subacute stroke patients.