I did another test. There is a normal SQLite table with a lot of town names. Here is the query: select * from LOCATION where loc_town REGEXP '[A-ZŁa-zł]'; Here is a fragment of the 6,000 rows in the result set: 52.534353 13.329043 80354118 8089118 BBEU Berlin Beusselstraße berlin beußelstrasse 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 47.811484 7.562759 80144154 8089119 RNBG Neuenburg Baden neuenburg baden 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 50.880221 6.085867 80151936 8089120 KXH Herzogenrath (grenze) herzogenrath grenze 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 51.736372 14.661882 80098558 8089122 BXFO Forst (grenze) forst grenze 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 53.415809 14.3734 80098509 8089123 WXG Grambow (grenze) grambow grenze 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 52.321174 14.576404 80098533 8089124 BXF Frankfurt Oder (grenze) frankfurt oder grenze 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 50.860058 14.22213 80098731 8089126 DXS Schöna (grenze) schöna grenze 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 53.325755 14.414598 80098517 8089127 WXT Tantow (grenze) tantow grenze 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 50.892428 14.829094 80098707 8089129 DXZ Zittau (grenze) zittau grenze 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 52.543253 13.368173 80319988 8089131 BWED Berlin Wedding berlin wedding 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 52.473101 13.455853 80354134 8089327 BSO Berlin Sonnenallee berlin sonnenallee 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 52.498738 13.269867 80354142 8089328 BMS Berlin Messe Süd Eichkamp berlin messe süd eichkamp 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 52.508123 13.259376 80030619 8089329 BHST Berlin Heerstraße berlin heerstraße 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 52.511026 13.241578 80030676 8089330 BOLS Berlin Olympiastadion berlin olympiastadion 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 52.510388 13.227132 80354159 8089331 BPIC Berlin Pichelsberg berlin pichelsberg 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 52.410761 13.308628 80354183 8089472 BLIS Berlin Lichterfelde Süd berlin lichterfelde süd 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 52.418842 13.314291 80198994 8089473 BOSS Berlin Osdorfer Straße berlin osdorfer straße 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 52.479375 13.352064 80030148 8089474 BSGR Berlin Schöneberg berlin schöneberg 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 52.486189 13.360927 80887562 8089537 BJLB Berlin Julius Leber Brücke berlin julius leber brücke 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 48.8659 8.509827 80117192 8090001 FFDF Ittersbach Rathaus ittersbach rathaus 80 0 2020-02-18 14:42:30 2020-02-18 14:42:30 48.608575 9.345716 80512285 8090021 TNU R Nürtingen Roßdorf There are a lot of rows in this set that should not have been part of the set. So lets try a different approach with a query that looks like this: select * from location where loc_town regexp '([A-ZŁ][a-zł])\w+'; Alas, the set still contains characters not defined by the regular expression. To avoid any discussion about Unicode and SQLite the database was defined with these settings: -- v0.21.01 New PRAGMA temp_store = MEMORY; PRAGMA synchronous = false; -- character encoding PRAGMA encoding = 'UTF-16le';