Agitated dispersions of one liquid in another immiscible liquid are widely used in chemical industry in operations such as liquid-liquid extraction, suspension polymerisation, and blending of polymers. When holdup of the dispersed phase is increased, in an effort to increase the productivity, at a critical holdup, the dispersed phase catastrophically becomes the continuous phase and vice versa. This phenomenon is known as phase inversion. Although the inversion phenomenon has been studied off and on over the past few decades, the mechanism of phase inversion (PI) has yet not become clear. These studies have however brought out many interesting aspects of PI, besides unravelling the effect of physical and operational variables on PL Experiments show that oil-in-water (o/w) and water-in-oil (w/o) dispersions behave very differently, e.g water drops in w/o dispersions contain oil droplets in them, but oil drops in o/w dispersions contain none, dispersed phase hold up at which inversion occurs increases with agitation speed for w/o dispersions but decreases for o/w dispersions. A common feature of both types of dispersions however is that as agitation speed is increased to high values, inversion holdups reach a constant value. A further increase in agitation speed does not change inversion hold up. Although this finding was first reported a long time ago, the implications it may have not received any attentions. In fact, the work reported in the literature since then does not even mention it. The present work shows that this finding has profound implications. Starting with the finding that at high agitation speed inversion hold up does not change with agitation speed, the present work shows that inversion hold up also does not change with agitator diameter, type of agitator and vessel diameter. In these experiments, carried out in agitated vessel, energy was introduced as a point source. The experiments carried out with turbulent flow in annular region of two coaxial cylinders, inner one rotating, in which energy is introduced nearly uniformly throughout the system, show that the inversion holdup remains unchanged. These results indicate that constant values of inversion holdups for a given liquid-liquid systems (o/w and w/o) are properties of the liquid-liquid systems alone, independent of geometrical and operational parameters. A new hypothesis is proposed to explain the new findings. Phase inversion is considered to occur as a result of imbalance between breakup and coalescence of drops. Electrolytes, which affect only coalescence of drops, were therefore added to the system to investigate the effect of altering coalescence of drops on phase inversion. The experiments performed in the presence of electrolyte KI at various concentrations indicate that addition of electrolyte increases the inversion holdup for both o/w and w/o dispersions for three types of systems: non polar-water, polar-water and immiscible organic-organic. Higher the concentration of electrolyte used, higher was the holdup required for phase inversion. These findings indicate that while the addition of electrolyte increases coalescence of drops in lean dispersions, it has exactly opposite effect on imbalance of breakage and coalescence of drops at high holdups near phase inversion point. The opposite effect of electrolytes in lean and concentrated dispersions could be explained qualitatively, but only in part in the light of a new theory, involving multi-particle interactions. The phase inversion phenomenon is quantified in a simple manner by testing the breakage and coalescence rate expressions available in literature. It has been found that, equilibrium drop size (where breakage and coalescence events are in dynamic equilibrium) approaches infinity near phase inversion holdup which is not an ex perimentally observed fact. To capture the catastrophic nature of phase inversion, two steady state approach is proposed. The two steady states namely the stable steady state and unstable steady state, are achieved by modifying the expression for coalescence frequency on the basis of (i) shear coalescence mechanism and, (ii) recognising the fact that at high dispersed phase holdup the droplets are already in contact with each other at all times and hence rendering the second order coales cence process to a first order one. Using two steady states approach, catastrophic phase inversion is shown to occur at finite drop size.