As more companies try to figure out how to monetize their online service offering, many struggle with whether to charge for subscriptions, or offer the services for free, with some kind of ad-supported model.
Here are a few things to ask when you are deciding whether to charge for online subscription

1.  Are people used to paying for this service? 

Companies that seem to have the most success charging subscriptions online tend to be replacing an offline service that consumers are used to paying for.  Some examples include entertainment (Rhapsody, Netflix, World of Warcraft), journalism (Barrons, New York Times–although these guys are struggling) and reference materials (Ancestry.com, Equifax Credit Watch, Encyclopedia Brittanica Online)

In these cases, the online subscription is simply a more convenient way of receiving existing services, or in some cases, an enhanced version of the services.  In general, if there's already a "budget line item" in the household ledger, people are more willing to pay.

2.  Does this service have a business appeal, or secondary market for small, or even medium (SMB) businesses?

Services like YouSendIt, Zoomerang and Phanfare can be useful for individual consumers, but also appeal to professionals (in these examples, creatives managing large files, researchers, and photographers respectively).  Services that help professionals get their work done are more likely to be worthy of subscription fees.  The more a service can prove it impacts your bottom line, the more likely you are to pay for it.  We are less willing to pay for things that make us better or more efficient, but don't actually save money. 

3.  Have you already been offering this same service for free?

Many great online services start out as free.  Entrepreneurs and VCs alike often focus first on building traction, showing that people really want to use the service, and then charging for the service once there is evidence that the service is useful.  Unfortunately, once you give me something for free, I feel entitled to continue getting it for free.  This is true whether it's free access to private land (like the Stanford "Dish" here in Silicon Valley) or access to an online social network, or video community. 

The exception to this rule is when you figure out a way to increase the value of what you have been offering, and price that service, keeping the original offering at free.  LinkedIn is a good example–their standard service continues to be free, but for $25-$500/month, you can get additional functionality.  LinkedIn is also a great example of tiered offerings at different prices, a key strategy of many successful subscription-based businesses.  In addition, LinkedIn makes good use of the freemium model, a term coined by New York VC Fred Wilson of Union Square Ventures.

Be careful when you start with a business model that has no financial component…some companies luck into a successful model, but most fail.

In general, consumer online services work best when the meet an existing need for which people are willing to pay, either as a private individual or as a business person.  But even the best of these services can shoot themselves in the foot by training their market to expect everything for free.