The aim of the study was to demonstrate the impact of different operational definitions (definition 1 -Amisco; definition 2 -Garganta, 1997; definition 3 -Garganta, 1997, adapted) over the frequency and location of ball recovery, and provide bases to select the most useful definition to an intended analysis. The sample is the first half of twelve matches from the Spanish soccer league. Matches were registered using AMISCO®. Data were examined using Cochran's Q test and Kruskal-Wallis test. Results showed that the frequency and location of ball recovery differ depending on the operational definition used. Definitions 1, 2 and 3 identified 909, 272 and 310 ball recoveries, respectively (p< 0.001 for all pairs of comparisons). The median distance from the goal to the location of ball recovery was, in meters, 40.0 (35.7) using definition 1, 32.2 (34.8) using definition 2 and 32.7 (34.3) when definition 3 was used (p<0.001 for comparisons between definition 1 and 2, and definition 1 and 3). The operational definitions impact in match analysis and consequently in the interpretation of the play. The criteria used should incorporate an instrumental selectivity that points to the purposes of the assessment. Concerning the frequency, definition 1 is the most glade, and definition 2 the most restrictive. Regarding the location, there were no significant differences between definition 2 and 3. Definition 2 allows to gain suitable and meaningful comparisons as previous studies has already based on it. Despite the differences, both definition 1 and 2 have a practical value. Ultimately, it is important to verify if the measurement instrument satisfies the logical intentions of the observational system .