The question whether controlled vocabulary or natural language free-text terms is the most effective search strategy has occupied researchers in LIS for many decades. This Masters thesis is an empirical study which aims to compare these two search strategies in LISA Library and Information Science Abstracts, an online bibliographic database. 22 topics from the discipline of Library and Information Science were constructed and out of each topic one query for each search strategy was formed. Queries in natural language were formed with terms from different sources, for example dictionaries, while queries in controlled vocabulary were built with terms from LISAs electronic thesaurus. The measures used in this study were precision, relative recall pooling method and overlap. The average precision was 65,59 % for controlled vocabulary while natural language generated an average precision of 41,75 %. Concerning the average relative recall, controlled vocabulary measured 40,28 %, while natural language reached 75,00 %. The average overlap was 15,28 %. Consequently, the controlled vocabulary accomplished a higher concentration of relevant documents but failed in finding several relevant documents. Natural language achieved a more exhaustive result but did also generate more non-relevant documents. The results suggest that these two strategies should be applied depending on situation and requirement but also that research on how to derive the advantages of both strategies in one search is needed.