Much interest has been generated in portable telephones and data terminals inside buildings. In a radio environment plagued by the severe fading effects of multipath and the dynamics of a changing channel, diversity is required. While considerable theory has been developed on diversity techniques, there have been few published results of indoor diversity measurements, particularly in the 1.7-2.0 GHz band. This paper evaluates three diversity techniques at 1.75 GHz based on simultaneous measurements of envelope fading due to both moving people and motion of a four-branch receiver. The experiment was conducted in various rooms on one floor of a university building having a common construction type of double plasterboard walls, cement floors, and acoustic tile ceilings. Path-loss characteristics of this environment had a distance power law exponent of N = 2.85 or equivalently, an estimated wall loss of 3.0 dB. Results presented here provide a direct comparison of two-branch space, two-branch frequency, and four-branch space/frequency (hybrid) diversity. Envelope cross correlations had averages less than 0.1 for space diversity with DELTA-S (antenna separation) > lambda/4 and values less than 0.2 for DELTA-F (frequency diversity spacing) > 10 MHz. Two-branch average diversity gain using selection combining was at least 10 dB at 99% signal availability for all DELTA-S including lambda/4 (4.3 cm), allowing such a technique to be used on a hand-held terminal, and for frequency diversity with DELTA-F > 5 MHz. With these parameters four-branch hybrid diversity had gains > 15 dB. The distribution of diversity gains varied within +/- 2.5 dB between transmission paths which should be considered for a worst-case link design. The reported gains allow reasonable estimations of system fade margins based on required availabilities when using diversity in similar buildings.