Most sources report the percentage of vocabulary in dictionaries, and the figures range from 53% to 70% - this will really depend on the dictionary used to calculate the figure.
The running words frequency is the percentage of words actually found in a text. This depends on the text used, but if a balanced representative corpus is used, you should get a fairly good estimate.
In 2002, the National Institute of Korean Language (국립국어원) published a study called 현대 국어 사용 빈도 조사 based on a corpus of 1,531,966 words, with the frequency of every word contained in the corpus. Based on that study, a newspaper article reports that the percentage of word origins among running words are as follows:
- Native Korean words ("토박이말"): 54%
- Hanja words (한자어): 35%
- Foreign words (외래어): 2%
I couldn't confirm those figures myself, as they don't seem to be in the original article published by the 국립국어원, but rather calculated from the tables provided in the article.