The importance of rivers for the human civilisation is great. Water always meant life. The human kind has a special interest in rivers for biological, economical and ecological reasons.

We can find representations of hydrography, along with relief and settlements on the earliest maps from Mesopotamia. These maps represent coastlines and rivers by lines as we do it today. A special and interesting problem is determination of river lengths.

Topographic maps are often used as a source for determination of river lengths. This method is time consuming and can sometimes yield results without required accuracy. The space for map content representation changes depending on map scale, and the number of details vary on maps of different scale. Loss of details is larger on maps at smaller scales. Additionally, it depends on the importance of the river, river curvature, as well as on line width used to represent the river on the map.

An analysis of river lengths on topographic maps of Croatia at different map scales is provided in this paper. A comparison with existing data is also given. The analysis was performed on 20 longest Croatian rivers. For practical reasons, only parts of rivers represented on Croatian topographic maps were taken into account.

River lengths (i.e. lengths of analysed lines) are shorter in smaller scales, but not as much as one might expect. The lengths determined on maps at scales 1:25 000 to 1:500 000 shows very small differences. These differences become more significant at scales of 1:1 000 000 and 1:2 000 000. The reason for that can be found in the fact that river lines are smooth enough and coastline details can be represented well on map in smaller scales.

Available data on river lengths are often given without information of methods or sources used. Therefore, a direct comparison of our results can’t be done. There are also differences in data from various sources, leading to confusion. A systematic and consistent approach of creating a database of river lengths is needed to serve as reference for such data.