The performance of a conically scanning satellite instrument for the measurement of cloud ice was studied. The instrument measures radiances in 12 channels placed around the 183, 325 and 448 GHz water vapour lines and the 243, 664 and 874 GHz window channels, and is designed to provide estimations of ice water path (IWP), the equivalent sphere diameter (DME), and the median ice mass height (ZME). Overall median relative errors of around 20% for IWP, 33 µm for DME, and 240 m for ZME for a midlatitude winter scenario, and 17% for IWP, 30 µm for DME, and 310 m for ZME for a tropical scenario were found. Detection limits (relative retrieval error reaching 100%) of around 2 gm -2 were estimated for both scenarios. The performance of a five-receiver instrument, where either the 664 or 874 GHz channel is dropped, was close, but with increased errors for very thin and high clouds. A trade-off between having the 874 GHz receiver or two infrared channels at 10.7 and 12 µm emerged, as very similar performance was found between the six-receiver instrument and the five-receiver instrument with the infrared channels. Another trade-off between receiver selection and noise was also apparent, with some of the four-receiver selections operating at half noise levels being able to compete with the standard six-receiver instrument. Dual-polarized measurements were also tested, but they did not significantly improve the retrievals of IWP or DME. ; The performance of a conically scanning satellite instrument for the measurement of cloud ice was studied. The instrument measures radiances in 12 channels placed around the 183, 325 and 448 GHz water vapour lines and the 243, 664 and 874 GHz window channels, and is designed to provide estimations of ice water path (IWP), the equivalent sphere diameter (DME), and the median ice mass height (ZME). Overall median relative errors of around 20% for IWP, 33 µm for DME, and 240 m for ZME for a midlatitude winter scenario, and 17% for IWP, 30 µm for DME, and 310 m for ZME for a tropical scenario were found. Detection limits (relative retrieval error reaching 100%) of around 2 gm -2 were estimated for both scenarios. The performance of a five-receiver instrument, where either the 664 or 874 GHz channel is dropped, was close, but with increased errors for very thin and high clouds. A trade-off between having the 874 GHz receiver or two infrared channels at 10.7 and 12 µm emerged, as very similar performance was found between the six-receiver instrument and the five-receiver instrument with the infrared channels. Another trade-off between receiver selection and noise was also apparent, with some of the four-receiver selections operating at half noise levels being able to compete with the standard six-receiver instrument. Dual-polarized measurements were also tested, but they did not significantly improve the retrievals of IWP or DME.