Farmers account for yield and soil variability to optimize their production under mainly economic considerations using the technology of precision farming. Therefore, understanding of the spatial variation of crop yield and crop yield development within arable fields is important for spatially variable management. Our aim was to classify landform units based on a digital elevation model, and to identify their impact on biomass development. Yield components were measured by harvesting spring barley (Hordeum vulgare, L.) in 1999, and winter rye (Secale cereale, L.) in 2000 and 2001, respectively, at 192 sampling points in a field in Saxony, Germany. The field was stratified into four landform units, i.e., shoulder, backslope, footslope and level. At each landform unit, a characteristic yield development could be observed. Spring barley grain yields were highest at the level positions with 6.7 t ha )1 and approximately 0.15 t ha )1 below that at shoulder and footslope positions in 1999. In 2000, winter rye harvest exhibited a reduction at backslope positions of around 0.2 t ha )1 as compared to the highest yield obtained again at level positions with 11.1 t ha )1 . The distribution of winter rye grain yield across the different landforms was completely different in 2001 from that observed in 2000. Winter rye showed the highest yields at shoulder positions with 11.1 t ha )1 , followed by the level position with 0.5 t ha )1 less grain yield. Different developments throughout the years were assumed to be due to soil water and meteorological conditions, as well as management history. Generally, crop yield differences of up to 0.7 t ha )1 were found between landform elements with appropriate consideration of the respective seasonal weather conditions. Landform analysis proved to be helpful in explaining variation in grain yield within the field between different years.