This thesis is motivated by the progressive expansion of electronic markets with reduced pre-trade transparency and the collateral liquidity effects. In this thesis I develop three independent theoretical models and explore the repercussions of weak market liquidity and transparency. First, I approach the issue of limited liquidity through the optimal order placement problem of a risk-averse trader in a continuous time context and introduce a random delay parameter, which defers limit order execution and characterises market liquidity. This framework demonstrates that imperfect liquidity explains order clustering in the proximity of best quotes and the existence of the bid-ask spread. The distribution of expected time-to-fill of limit orders conforms to the empirically observed distribution of trading times, and its variance decreases with liquidity. Finally, two additional stylised facts are rationalised in this model: the equilibrium bid-ask spread decreases with liquidity, but increases with agents’ risk aversion. My second framework adjoins the few theoretical attempts in the literature to challenge investors’ incentives to participate in opaque trading environments. Through a real option approach I justify how market opacity can encourage liquidity provision, which, in turn, supports the empirical evidence on the proliferation of such trading venues. I demonstrate that transparency in conjunction with liquidity determine traders’ eagerness to supply liquidity to the opaque market, and that once the trader enters the opaque market, he commits to trade relatively quickly. Furthermore, error analysis reveals that impatient traders are highly likely to pass over favourable trade execution offered in the opaque market precisely because of imperfect clarity of information signals, while a prior optimistic bias prompts the trader to submit his limit order sooner. Lastly, as a complement to static inference on individual level, I establish an artificial market with heterogeneous agents and distinct transparency regimes that replicates the long memory properties and empirical order flow patterns. The results suggest that full quote transparency incurs substantial transaction costs and dampens trading activity, while exogenous restriction of displayed depth up to several quotes does not alter significantly market performance. The core implication of this model is that the endogenous restriction of displayed quote depth by means of iceberg orders improves market quality in multiple dimensions. This thesis contributes to the microstructure domain by providing a theoretical support to the benefits of market opacity as well as its downsides. The research outcomes of this thesis are, therefore, relevant from a regulatory standpoint.