The recent availability of large data sets covering single transactions on financial markets has created a new branch of econometrics which has opened up a new door of looking at the microstructure of financial markets and its dynamics. The specific nature of transaction data such as the randomness of arrival times of trades, the discreteness of price jumps and significant intraday seasonalities, call for specific econometric tools combining both time series techniques as well as microeconomtric techniques arising from discrete choice analysis. This paper serves as an introduction to the econometrics of transaction data. We survey the state of the art and discuss its pitfalls and opportunities. Special emphasis is given to the analysis of the properties of data from various assets and trading mechanisms. We show that some characteristics of the transaction price process such as the dynamics of intertrade durations are quite similar across various assets with different liquidity and regardless whether an asset is traded electronically or on the floor. However, the analysis of other characteristics of transaction prices process such as volatility requires a careful choice of the appropriate econometric tool. Empirical evidence is presented using examples from stocks traded electronically and on the floor at the German Stock exchange and from BUND future trading at the LIFFE and the EUREX.