Compact hardware architectures are proposed for the ISO/IEC 10118-3 standard hash function Whirlpool. In order to reduce the circuit area, the 512-bit function block ρ[k] for the main datapath is divided into smaller sub-blocks with 256-, 128-, or 64-bit buses, and the sub-blocks are used iteratively. Six architectures are designed by combining the three different datapath widths and two data scheduling techniques: interleave and pipeline. The six architectures in conjunction with three different types of S-box were synthesized using a 90-nm CMOS standard cell library, with two optimization options: size and speed. A total of 18 implementations were obtained, and their performances were compared with conventional designs using the same standard cell library. The highest hardware efficiency (defined by throughput per gate) of 372.3 Kbps/gate was achieved by the proposed pipeline architecture with the 256-bit datapath optimized for speed. The interleaved architecture with the 64-bit datapath optimized for size showed the smallest size of 13.6 Kgates, which requires only 46% of the resources of the conventional compact architecture.