Studies show that reading is one of the most effective ways of vocabulary-learning, no matter whether you have been exposed to the target language or not. It has been proved by some researchers that L2 learners may acquire the meanings of novel words from sentence context while reading. But neither have they compared the difference of subjects’ performance between different types of sentence length nor different types of context constraint, and therefore their works left us with questions. As a matter of fact, context constraint is such a powerful factor that it can has great impacts on semantic and syntactic processing in sentence comprehension according to some studies. We propose that both context constraint and sentence length are important factors in indirect vocabulary learning. Furthermore, Working memory has been considered as an important part in some advanced cognitive processes, such as: reading, comprehension, reasoning, etc. Some studies do find some relationship between working memory capacity and L2 word knowledge, language production or vocabulary semantic extraction, but there is still no study focusing on the role of working memory while learning words via reading. In this study, we presented an experiment to explore whether working memory capacity, sentence length and context constraint affect the performance of vocabulary learning while reading in L2, and how exactly they work. We tested three different age groups with average ages of 14, 17 and 21 respectively. All subjects were required to finish a vocabulary-learning task after the working memory operation span task. We made reading materials differing in 3 types of sentence length and 2 types of context constraint, and divided the subjects of each age group into two groups according to their working memory span scores. For those who come from the middle or the eldest age group, the results show that working memory capacity do have a significant relationship with their performance in the vocabulary-learning task. While in the condition of mid-length and long sentence, subjects got significant higher scores in the vocabulary-learning task when the context was restricted much more strictly. And in contrast to subjects of the youngest group, the longer the sentence mid and elder aged subjects read, the worse they did in the vocabulary-learning task when the context was not that restricted. Evidence of this study offered strong support for the hypothesis that working memory plays an important role in indirect vocabulary learning and its impact won’t be diminished no matter what kind of reading materials is involved. In other words, we may make vocabulary learning much easier via working-memory-related training. Additionally, our findings in reading materials suggest controlling sentence length and context constraint of study materials can also achieve the same effect. More specifically, the more strictly context was restricted, the more efficient L2 vocabulary learning can be.