Many public API's (twitter, for example) return a collection of json objects for a request for a collection of resources. The mongoimport tool only understands newline-delimited json objects in a file. In other words, it only understands one json object at a time. It seems like mongoimport should know how to understand these collections natively. For example, trying the following will result in an error:

curl http://api.twitter.com//1/statuses/user_timeline.json?screen_name=noradio > noradio.twitter.json
mongoimport -d twitter -c noradio --file=noradio.twitter.json