Recently, a vast number of scientific publications have been produced in cities in emerging countries. It has long been observed that the publication output of Beijing has exceeded that of any other city in the world, including such leading centres of science as Boston, New York, London, Paris, and Tokyo. Researchers have suggested that, instead of focusing on cities' total publication output, the quality of the output in terms of the number of highly cited papers should be examined. However, in the period from 2014 to 2016, Beijing produced as many highly cited papers as Boston, London, or New York. In this paper, I propose another method to measure cities' publishing performance; I focus on cities' publishing efficiency (i.e., the ratio of highly cited articles to all articles produced in that city). First, I rank 554 cities based on their publishing efficiency, then I reveal some general factors influencing cities' publishing efficiency. The general factors examined in this paper are as follows: the linguistic environment, cities' economic development level, the location of excellent organisations, cities' international collaboration patterns, and the productivity of scientific disciplines.