This chapter focuses on the statistical analysis of lexico-grammatical features in language (such as articles, passive constructions or modal expressions). We start with a discussion of two types of approaches to lexico-grammar in corpora. The first approach uses the ‘Whole corpus’ research design and compares the frequencies of a linguistic variable (and its variants) in broadly defined subcorpora. The second approach employs the ‘Linguistic feature’ research design and carefully defines the contexts in which a particular variable can occur (i.e. its lexico-grammatical frame) and analyses factors which contribute to the occurrence of one variant of the variable as opposed to another. Following the second approach, the chapter shows how lexico-grammatical variation can be summarised using cross-tabulation and what statistical measures can be computed based on cross-tabulation summary tables. These measures range from simple percentages to the chi-squared test and logistic regression. Since logistic regression represents an advanced statistical procedure, large parts of the chapter are devoted to explaining this method and the interpretation of its output.