We present the data reduction pipeline for the Herschel Infrared\ud
Galactic Plane survey (Hi-GAL). Hi-GAL is a key project of the Herschel\ud
satellite, which is mapping the inner part of the Galactic plane\ud
(vertical bar l vertical bar <= 70 degrees and vertical bar b vertical\ud
bar <= 1 degrees), using two Photodetector Array Camera and Spectrometer\ud
(PACS) and three Spectral and Photometric Imaging Receiver (SPIRE)\ud
frequency bands, from 70 to 500 mu m. Our pipeline relies only partially\ud
on the Herschel Interactive Processing Environment (HIPE). It features\ud
several newly developed routines to perform data reduction, including\ud
accurate data culling, noise estimation and minimum variance map-making,\ud
the latter performed with the ROMAGAL algorithm, a deep modification of\ud
the ROMA code already tested on cosmological surveys. We discuss in\ud
depth the properties of the Hi-GAL science demonstration phase data