Well I guess the main difference is that germ line mutations can be transmitted to offspring - either in a autosomal dominant or a recessive manner. Somatic mutations cannot and (presumably) were caused by a de novo mutation in the somatic tissue.
An example of a somatic mutation may include mutations that lead to cancer in the adult, an example of a germ line mutation may be Duchenne muscular dystrophy.
Obviously germ line mutations can be de novo too - in which case they are unlikely to generate a phenotype in the person carrying the mutation but are likely to manifest in the offspring only.
It is not an inherited condition if it is only found in somatic cells. Since you can't measure the affected pt's germline the standard strategy is to look for the mutation in the parents. But not finding it in the parents does not rule out the possibility of a new mutation the pt's germline. "A new disease mutation" that is only in somatic cells requires proof that this is not a sporadic mutation only found in this individual's tissue. If you have a somatic mutation only found in one individual it would be overreaching to call this a "new disease". Other cases would likely be necessary though the finding is important and may be publishable.
We have found several cases in which the mother of a patient with a X-linked disorder is showing the mutation in 4%, 8% and 16% of her DNA resp. So this means that it is possible to also have a mosaic pattern that is present in all tissues. We tested DNA from buccal swaps and DNA purified from blood. So you should check multiple tissues for the mutation.
Germ line mutations can be of great clinical importance in various human diseases, for example in cancer. Mutations in these genes may predispose a subject to a particular form of cancer depending on type of gene and mutation involved. In such cases these genes are called as cancer predisposing genes (CPGs). CPGs can be used for early diagnosis by personalized DNA sequencing. Some CPGs also predict therapeutic efficacy so these can be used to decide the course and type of treatment. More that 110 CPGs have been identified with links to cancer.
Another point is that presence of a mutation does not really mean that it can be linked to a particular disease condition. Some mutations are rare and hence we can find them in rare cases when profiling molecular basis of cancer. But it does not mean that presence of that mutation is responsible for cancer.
Germline mutations can be constitutional mutation that is present in all cell types. So I agree with Martin that you have to check different types of cell. Try to check it on germ cells.