We theoretically develop and empirically study the concept of geographic representation. We identify an overlooked aspect of (mis)representation in electoral systems: The degree to which within parties, voters residing in some districts are overrepresented at the expense of their co-partisans residing elsewhere. Drawing on district-level data of 113 parties from 12 districted democracies, we find a substantial degree of geographic representational discrepancies within parties. Importantly, we show that the discrepancy is often accompanied by a difference in policy positions between under and overrepresented districts, rendering it particularly significant politically. We derive a general measure for the discrepancy between the votes a party gains and the seats it holds in different districts, namely, geographic disproportionality (GeoDisp). Unlike malapportionment, GeoDisp is party specific, and draws on actual votes. Utilizing geocoded data, we find a substantial degree of geographic disproportionality, which, per our expectations, varies by party.