A dimorphic yeast Kluyveromyces marxianu var. marxianus NRRLy2415 which exhibits a wide range of mean morphological forms was used as a model organism to investigate the role of cell morphology on the dead-end and crossflow filtration behaviour. Varying the culturing conditions produced cell suspensions of different mean morphology. Batch fermentations were used to produce yeast-like morphologies and continuous cultures produced cells more mycelial in nature. Semi-automated image analysis was used to measure the mean specific surface area, Sv, and the mean ratio of cell length to equivalent cylindrical diameter, ¿dm- The parameter L&m was used to describe the mean cell morphology, whereby increasing values of Ldm indicate that the mean cell morphology is more mycelial in nature. It was shown that a linear relationship was found to exist between the mean specific resistance, a av, and applied pressure, AP, over the entire range of pressures employed and for all morphologies of K. marxianus filtered. The linear relationship, i.e. a av = a 0(l + A:cAP) allowed the null stress resistance, a 0, to be determined, thus uncoupling cake packing and compressibility effects. The voidage of cell pellets formed in a bench-top centrifuge was measured for a range of cell suspensions of different mean morphologies, and in all cases, the pellet voidage decreased linearly with centrifugation speed. This decrease became more pronounced as L(\m increased. The extrapolated null stress pellet porosity, s0, was correlated with ¿dm by the expression s0= l -1/(1.24+0.0621dm)- The ratio 0-opc/ S ] decreased with increasing Ldm, a finding that was qualitatively consistent with the pellet voidage data and the Kozeny-Carman equation. Considerably better agreement with the experimental data was obtained when the Kozeny constant, k0, was treated as a variable and related to Zdm by the equation k0 = 10.1 + 0.34Zdm. It was shown that as L{im increases that filter cake compressibility defined as kc increases. The reversibility of filter cake compression was determined by measuring the specific resistance of a pre-formed filter cake, exposing the cake to a higher pressure and then re-determining a av at the lower pressure. It was found that the compression of filter cakes of K. marxianus and bakers yeast was predominantly reversible, whereas with filter cakes of calcium carbonate, compression was completely irreversible. A model was presented to qualitatively describe filter cake compression. The model was found to predict a near linear relationship between a av and AP, as was found to occur with K. marxianus and other yeast suspensions. It was also shown that the Tiller relationship relating a av and AP could be reduced to a linear expression if the index of compressibility, n \, is equal to 2. Furthermore it was shown theoretically that the coefficient of compressibility increases with null stress porosity, thus partially explaining the increase in kc with Ldm. Crossflow filtration studies were conducted with a tubular carbosepharose membrane. In crossflow filtration experiments, cake formation, fouling and evolution of mean specific resistance of suspensions of K. marxianus was studied. Suspensions washed in saline and unwashed suspensions of K. marxianus grown in different media and of various mean morphologies, were used to analyse cake formation and fouling. It was shown that fouling is significant during crossflow filtration whilst it is negligible during dead-end filtration. Varying the cell morphology influenced the filtration behaviour more than using different growth media for both washed and unwashed cell suspensions. Spent media separated from the cells by dead-end filtration was used to study the effect of media components on fouling. The filtration flux of spent medium was found to be independent of the growth conditions and only slightly influenced by the culturing media. For unwashed suspensions there appeared to be considerable rejection of macrosolutes during crossflow filtration that resulted in fluxes an order of magnitude lower than for washed suspensions. By switching the feed to saline solution during the pseudo steady state period it was shown that the cell cake remains largely intact and thus is not significantly influenced by shear. A novel technique to determine cake mass during crossflow filtration was developed. The cake mass was determined by conducting dry weight analysis on samples withdrawn from the feed reservoir during filtration and subsequently doing a mass balance on the system. For the washed cell suspensions, it was shown that with increasing L&m, the mass of cake formed decreased from approximately 6g to 2g. Furthermore preferential deposition of small cells does not appear to occur with washed yeast-like suspensions as indicated by the cake formation rate and image analysis data. However, preferential deposition of smaller cells was found to occur with unwashed yeast suspensions and possibly occurred with both washed and unwashed mycelial suspensions. Recovery of filter cakes comprised of yeast cells was significantly higher than with mycelial filter cakes, indicating possible differences in cake structure and cell membrane interactions. The specific resistance of washed cell suspensions increases throughout filtration and is higher than observed during dead-end filtration even if membrane fouling is considered. This indicated a degree of cake fouling by fines occurring during crossflow filtration. Also the specific resistances of the recovered crossflow filter cakes, measured by dead-end filtration, were significantly greater than those for the feed suspensions. The apparent specific resistances of mycelial filter cakes were greater than those determined for yeast morphologies. This is in contrast to deadend filtration where the specific resistance was less for mycelial suspensions than for yeast suspensions. The effect of varying the crossflow velocity with washed yeast-like suspensions was studied. It was shown that the cake mass decreased and filtrate flux increased with increasing crossflow velocity. Furthermore it was shown the specific resistance decreases with crossflow velocity. It was shown that this was probably attributable to the pressure drop across the filter cake decreasing due to decreasing cake mass, and hence less cake compression. There was also evidence of fouling decreasing with increasing crossflow velocity.