Following the suggestions by Jonathan Druart, Mark Tompsett and Barton Chittenden the problem is solved by disabling the zebra_facets option in the koha-conf.xml file. The next question is about what Barton Chittenden says: when some words like articles or prepositons and included in the query string, in one hand the search is slower and on the other hand the amount of "noise"* in the response is great. In fact the most problematics searches where those in which the spanish word "de" (of) appeared. What could be the solution for this problema? Disable the Fuzzy and Stemming search? I read that there is no stopwords list in Koha today. Could I write a function for "cleaning" the query string before the seach is made? Thanks in advance. Saludos Camilo *I dont know if its the suitable english word...