Root-soil relationships are pivotal to understanding crop growth and function in a changing environment. Plant root systems are difficult to measure and remain understudied relative to above ground responses. High variation among field samples often leads to non-significance when standard statistics are employed. The adaptive neuro-fuzzy inference system (ANFIS) has been applied in many agricultural and environmental fields and may represent a viable means for dealing with complexities of root distribution in soil. We applied this method to vertical and horizontal root distribution data collected from a potato (Solanum tuberosum L.) cropping system grown under ambient and elevated levels of atmospheric CO2. The lack of a CO2 effect on root length or dry mass densities was most likely due to the low growing season temperature limiting root growth in this subarctic system. At all CO2 levels, potato roots were concentrated near row centre, particularly in the upper soil profile. Simulations indicated that ANFIS gave plausible results, indicating it offers a viable alternative to more traditional statistical techniques for evaluation of complex root distribution patterns.