In simple-flow shop problems, each machine operation center has one machine. If a machine center has more than one machine, scheduling becomes a flexible shop flow problem. Flexible shop flows are thus generalizations of simple shop flow. Processing time for each job was previously assumed to be known exactly, but in many real-world applications, processing times vary dynamically due to human or operating factors. We apply fuzzy concepts to represent uncertainty in processing times for flexible-flow shops with two machine centers, then propose a fuzzy heuristic algorithm for job scheduling. We first use a fuzzy LPT algorithm to allocate jobs, then use a fuzzy Johnson algorithm to deal with task sequencing. The proposed method lets jobs be scheduled more flexibly than conventionally.