A method and apparatus for efficient memory bank utilization in multi-threaded packet processors is presented. A plurality of memory access requests, are received and are buffered by a plurality of memory First In First Out (FIFO) buffers, each of the memory FIFO buffers in communication with a memory controller. The memory access requests are distributed evenly across said memory banks by way of the memory controller. This reduces and/or eliminates memory latency which can occur when sequential memory operations are performed on the same memory bank.