This work describes different strategies for the detection of biologically relevant analytes. The main goal was to achieve sensing in water, which is a prerequisite for application to real biological samples. In several cases, we could demonstrate that our systems were also working in blood serum, a notoriously complex medium containing proteins, sugars and salts, which can potentially interfere with our sensors. Our first analyte of choice was the lithium cation. Owing to its therapeutic uses in the field of psychology, it represents an appealing target, and an efficient chemosensor for lithium ions may find numerous uses in biomedical investigations. The first strategy that will be presented involves the use of organometallic macrocycles carrying fluorescent reporting units. These compounds were found to be potent ionophores and display a very high selectivity for lithium ions. Upon careful choice of constituent building blocks, a chemosensor displaying good affinity, selectivity, and solubility was obtained. The sensor allowed the detection of lithium ions in water and in human serum by simple fluorescence measurements (Chapter 2). Other approaches for the detection of lithium ions will be presented in Chapter 3. Instead of covalently attaching fluorescent dyes to the lithium binding unit, fluorophores that can bind in a non-covalent fashion were employed. Assays based on easily available constituents were devised. They allowed the sensing of lithium ions in the sub-millimolar concentration range. In Chapter 4, the pattern-based sensing of micromolar concentrations of small peptides in water will be described. Small peptides are ubiquitous in biology, and they are involved in many essential processes such as neurotransmission, blood pressure regulation or oxidative stress mitigation. It will be demonstrated that their selective detection can be accomplished by arrays of differential sensors: instead of building a selective sensor for a given analyte, a series of cross-reactive, unsophisticated sensors could be employed. Their differential response (fingerprint) for various analytes was interpreted via statistical methods. As a result, low micromolar concentrations of many short peptides (as well as mixtures thereof) could be measured in water and in human serum samples. In the last part (Chapter 5), the sensing properties of a sensor array will be compared with those of a dynamic combinatorial library (DCL). The DCL contains the same elements as the sensor array, but they are mixed in one pot. It is shown that the DCL can outperform the sensor array in many situations, but that an excessive complexity can be detrimental to its resolution.