The density, workability, porosity, strength development, and durability of gypsum-Portland cement-natural pozzolan blends are discussed in this paper. The results indicate that the variation in proportions of the mix constituents causes changes in the density. The effects of adding a superplasticizer depend on the amount of added. The higher dosages of a naphthalene-based superplasticizer pen-nit a much greater increase of workability. This improvement in workability can be used to produce mixes with a lower water/binder ratio while maintaining constant workability. The gypsum-Portland cement-natural pozzolan blend gives lower strength than gypsum at age of 1 day. However, its strength continues to increase from ( day onward and exceeds that of gypsum at 28 days. The effect of porosity on the strength can be described with a quadratic exponent for gypsum-Portland cement-natural pozzolan blends. The porosity tends to decrease significantly with the curing period. The rate of reduction is high for curing periods up to 28 days. However, above this period, the rate is much reduced. The durability assessments are made by examining the behaviour of gypsum-Portland cement-natural pozzolan blends in water and freezing and thawing cycles. The blends with compositions of 41:41:18 (gypsum/Portland cement/natural pozzolan) and 41:41:1851 (gypsum/Portland cement/natural pozzolan/1% superplasticizer) give excellent property retention after ageing in water at 20 degreesC for 95 days. An exponential equation sigma=sigma[1+(l -e(-bt))/a] where sigma(o) is the initial strength of gypsum-Portland cement-natural pozzolan blend, sigma is the strength of the gypsum-Portland cement-natural pozzolan blend after exposure to the water for time t, t is the duration of exposure to the water, and a and h are the material constants to be determined, is proposed to predict the variation of compressive strength of these blends immersed in water with time. This equation shows excellent agreement with the data on 41:41:18 and 41:41:18S1 in a period of 95 days. These blends perform very well during the freezing and thawing test. They do not show any deterioration up to 20 cycles of freezing and thawing. However, the test results after 30 cycles show that the scaling resistance of 41:41:18Sl is somewhat better than that of 41:41:18. (C) 2002 Elsevier Science Ltd. All rights reserved.